climas...climas • institute for the study of planet earth climate assessment for the southwest...

2001 Fire & ClimateWorkshops

Workshop Proceedings

February 14-16, 2001

March 28, 2001

Tucson, Arizona

edited by

Gregg Garfin and

Barbara Morehouse

CLIMAS• Institute for the Study of Planet Earth

Climate Assessment for the SouthwestClimate Assessment for the Southwest

2001 Fire & Climate Workshops

Workshop Proceedings

February 14-16, 2001March 28, 2001

Tucson, Arizona

Edited byGregg Garfin, Assistant Staff ScientistInstitute for the Study of Planet Earth

Barbara Morehouse, Program ManagerInstitute for the Study of Planet Earth

Published byThe Climate Assessment Project for the Southwest (CLIMAS)Institute for the Study of Planet EarthThe University of ArizonaTucson, Arizona

September 2001

ii

We are grateful to the following institutions and individuals for generously providing funding for the2001 Fire & Climate Workshops.

National Interagency Fire CenterThe Joint Fire Science Program ........................................................................................................ Bob Clark

National Oceanic and Atmospheric Administration (NOAA)Office of Global Programs ....................................................................................................... Roger Pulwarty

United States Forest ServiceRiverside Fire Laboratory .......................................................................................................... Francis Fujioka

University of ArizonaCLIMAS (funded by the NOAA Office of Global Programs) ............................................. Barbara MorehouseInstitute for the Study of Planet Earth................................................................................ Jonathan OverpeckLaboratory of Tree-Ring Research .............................................................................................. Tom Swetnam

The workshops were developed and hosted by the following institutions and individuals:

Tim Brown, Program for Climate, Ecosystem and Fire Applications (CEFA), Desert Research InstituteGregg Garfin, Climate Assessment for the Southwest (CLIMAS), University of ArizonaBarbara Morehouse, Climate Assessment for the Southwest (CLIMAS), University of ArizonaJonathan Overpeck, Institute for the Study of Planet Earth, University of ArizonaTom Swetnam, Laboratory of Tree-Ring Research, University of Arizona

Publication of this proceedings was sponsored by:

CLIMAS, Climate Assessment for the Southwest, University of ArizonaThe Joint Fire Science Program, National Interagency Fire CenterInstitute for the Study of Planet Earth, University of Arizona

Sponsors

iii


Foreword

Welcome to the printed version of the presentations anddiscussions of the 2001 fire-climate workshops, Fire andClimate 2001 (convened February 14-16, 2001) andFire and Climate in the Southwest 2001 (convenedMarch 28, 2001). The information contained hereinconsists of illustrated summaries of the presentationsgiven at the workshops. All presentations, with the ex-ception of the introductory and logistical remarks, arerepresented. Discussion summaries are limited to syn-theses of the breakout group presentations, as these rep-resent the culmination of formal and informal discus-sions that occurred throughout the workshops. A list ofattendees is provided at the end of this volume.

As this volume contains presentations from both fire-climate workshops, we have presented the material inthe order in which the presentations were given at eachworkshop. In order to make this proceedings volumeeasy to read and to put the presentations in context,workshop agendas have been included, and the tablecontents has been arranged in two ways, by order ofpresentation and by presenter.

The climatic backdrop for convening of these work-shops in the spring of 2001 was not as spectacular orcertain as the persistent and strong La Niña conditionsthat prompted our February 2000 workshop, The Im-plications of La Niña and El Niño for Fire Management.However, the smoldering aftermath of the devastating2000 fire season, provided a strong impetus for us tobring together for a second time fire managers, climatescientists, and fire researchers, in order to discuss therole of climate in wildfire regimes, and the needs of firemanagement professionals for climate information.Moreover, the very uncertainty accompanying the win-ter-spring 2001 climate forecasts, as we transitionedout of La Niña conditions, underscored the need forfurther dialogue: for fire managers to better under-stand climate forecast tools, their correct interpretationand their limitations, and for climate forecasters tobetter understand the decision support needs of firemanagement professionals and the constraints underwhich they need to make decisions that affect ecosys-tem health, property and human life.

Fire and Climate 2001. Due to the fact that resourceallocation for fire management requires coordinated

decision-making at regional and national levels, theability to make accurate climate forecasts for any partof the country has ramifications for resource allocationnationwide. The experience of the blazing summer of2000 highlighted the need for the climate/weather andwildfire/land management communities to work to-ward a more integrated understanding of the role ofclimate, as well as more immediate weather, in wildfireoccurrence and land management decision-making.Thus, we invited regional-level representatives of thefire management, climate science, integrated assess-ment, and fire research communities from much of thecontinental United States and Alaska to attend Fireand Climate 2001. For two and one-half days, we dis-cussed the linkages between fire, long and short-termclimate variations, communication and interpretationof long-range climate forecasts and forecast uncer-tainty, and user-driven integrated climate assessmentand climate service initiatives. We developed a set ofrecommendations for promoting education about cli-mate for fire managers, as well as fostering communi-cation between climatologists and fire managers.Moreover, we identified areas of research that will fa-cilitate better management decisions with regard to theconfluence of fire, climate and society.

Fire and Climate in the Southwest 2001. The South-west, home to CLIMAS and the University of Arizona,is a region of special focus. The arid foresummer andsummer monsoon, characterized by intense, spatiallyscattered, downpours and spectacular fire-igniting py-rotechnics, provide a unique set of circumstances forfire management. Rapid population growth in theSouthwest and increased development along the out-skirts of major cities, has accelerated the trend towardconflation of urban and rural land use patterns and as-sociated increase in fire risk at the urban-wildland in-terface. The escaped prescribed fire at Bandelier Na-tional Monument, accelerated by prolonged droughtand locally intense winds, which threatened LosAlamos during the spring of 2000, brought the issue ofclimate-fire-society relationships into sharper focus. Aspart of CLIMAS’ ongoing commitment to dialogue,communication, and collaboration with decision mak-ers in our region, we invited representatives of the Ari-zona and New Mexico fire management and climatescience communities to Tucson for the one-day meet-

iv

ing, Fire and Climate in the Southwest 2001. Fire man-agers were given a whirlwind tour of Southwest cli-mate, historical climate-fire patterns, and climate fore-cast skill, uncertainty and interpretation. At the meet-ing, we discussed the needs of the Southwest fire man-agement community for climate information and edu-cation, and we developed a set of recommendations forfire-climate research, decision-support tools and waysto negotiate institutional constraints with regard to thespecific needs of Southwest fire managers.

In retrospect, the meetings were highly successful.Workshop participants were enthusiastic and the rec-ommendations developed during each of the meetingshave formed the core for research proposals, traininginitiatives and continued communication between theclimate science and fire management communities.Moreover, workshop evaluations and breakout grouprecommendations provided valuable informationabout the direction of future workshops.

We thank the participants for the contribution of theirexpertise and their enthusiastic participation. We offerspecial thanks to the presenters for having taken timefrom their busy schedules to share their expertise andexperience.

Tim Brown, Director, CEFAGregg Garfin, Assistant Staff Scientist, CLIMASBarbara Morehouse, Program Manager, CLIMASJonathan Overpeck, Director, ISPEThomas Swetnam, Director, LTRR

Note: Many of the figures in this

volume are best viewed in color. You

can download the full-color version of

the proceedings from our wesite:

http://www.ispe.arizona.edu/climas/fire/

v


Acknowledgments

The following individuals provided valuable input to developing workshop content.

Richard Okulski, National Weather Service, TucsonKelly Redmond, Western Regional Climate Center, Desert Research Institute

As with all such endeavors, considerable effort, creativity, and good humor were required to make these workshopsa success. Special thanks go to the following individuals for their assistance. We apologize if we have left anyone’sname off the list.

Rex Adams, Education and Outreach Specialist, LTRRChris Basaldu, Research Assistant, CLIMAS/Dept. of AnthropologyDave Brown, Research Assistant, CLIMAS/Dept. of GeographyDon Falk, Research Assistant, LTRRCalvin Farris, Research Assistant, LTRRCaroline Garcia, Business Manager, Arizona Research LabsDerek Honeyman, Research Assistant, CLIMAS/Dept. of AnthropologyStacey Hines-Holdcraft, Administrative Associate, ISPESandy Jacobson, Administrative Assistant, ISPEKorine Kolivras, Research Assistant, CLIMAS/Dept. of GeographyEllis Margolis, Research Assistant, LTRRAna Martinez, Administrative Assistant, LTRRShoshana Mayden, Information Specialist Coordinator, ISPEStephen A. McElroy, USDA Agricultural Research ServiceCarmen Melendez-Allen, USFS, Coronado National ForestJane Moody, Research Assistant, CLIMAS/Dept. of AnthropologyPhyllis Norton, Administrative Associate, LTRRErika Trigoso, Research Assistant, CLIMAS/Dept. of AnthropologyColin West, Research Assistant, CLIMAS/Dept. of AnthropologyTamara Wilson, Research Assistant, CLIMAS/Dept. of GeographyBarbara Wolf, Research Assistant, CLIMAS/Dept. of Anthropology

1


Table of Contents

Weather, Climate, Fire: The 2000 Fire Season and Beyond ......................................... 3

Climate and Fire: Framing the Issues in Context of the 2000 Fire Season Tim Brown ................................... 3Weather and the 2000 Fire Season Rick Ochoa .............................................................................................. 7NIFC Resource Impacts During the 2000 Fire Season Rick Ochoa .............................................................. 8Climate Forecasts for 2000 and 2001 Klaus Wolter ........................................................................................ 9Fire Forecasts for 2001 Tim Brown ............................................................................................................. 12Recent Developments in Data Access at Western Regional Climate Center Kelly Redmond.......................... 15The Role of Weather and Climate at Cerro Grande: A Quick Overview John Snook .................................... 17

Beyond the 2000 Fire Season .................................................................................................... 20

MAPSS: Mapped Atmosphere Plant Soil System Ron Neilson ...................................................................... 20An Overview of the Joint Fire Science Program and RFPs for 2001 Bob Clark ............................................ 21

Climate Prediction .......................................................................................................................... 22

Climate Modeling Overview and Climate Forecast for 2001 Klaus Wolter ................................................... 22Seasonal Climate Forecasts: Wildfire Applications Dan Cayan .................................................................... 24Climatology of Western Wildfire and Experimental Long-Range Forecasts of Wildfire Season Severity Anthony Westerling ..................................................................................................................... 25Climate Prediction: ENSO vs Non-ENSO Conditions Douglas LeComte .................................................... 29Climate Prediction Interpretation and Evaluation Workshop Holly Hartmann and Tom Pagano ................. 33

Integrated Assessments ............................................................................................................... 40

Interagency Integrated Assessments Roger Pulwarty ..................................................................................... 40Climate and Human Impacts on Fire Regimes in Forests and Grasslands of the U.S. Southwest Barbara Morehouse and Steve Yool ............................................................................................................ 41Program for Climate, Ecosystem and Fire Applications Overview Tim Brown ............................................ 44Climate, Fire, and the Need for a National Climate Service Jonathan Overpeck ........................................... 47

Fire and Climate in the Southwest 2001 .............................................................................. 49

Climate of the Southwest Andrew Comrie ................................................................................................... 49Climate and the 2000 Fire Season in the Southwest Tom Swetnam .............................................................. 52Fire and Emergency Management in Arizona Alex McCord ......................................................................... 55A Brief Introduction to Climate Forecast Resources for Fire Management Ron Melcher ............................... 55Fire History in the Jemez Mountains: The Bandelier Perspective L. Dean Clark .......................................... 57Climate Forecasts Overview for the Southwest Tom Pagano and Holly Hartmann ........................................ 58Breakout Groups ......................................................................................................................................... 61

Appendix

Participant List ............................................................................................................................................. 66Agenda: Fire and Climate 2001(February) ................................................................................................... 74Agenda: Fire and Climate in the Southwest 2001(March) ............................................................................ 75

2

List of Presenters

Tim Brown Climate and Fire: Framing the Issues in Context of the 2000 Fire Season ................... 3Fire Forecasts for 2001 .............................................................................................. 12Program for Climate, Ecosystem and Fire Applications Overview ............................. 44

Dan Cayan Seasonal Climate Forecasts: Wildfire Applications ..................................................... 24

Bob Clark An Overview of the Joint Fire Science Program and RFPs for 2001 .......................... 21

L. Dean Clark Fire History in the Jemez Mountains: The Bandelier Perspective ............................... 57

Andrew Comrie Climate of the Southwest .......................................................................................... 49

Holly Hartmann Climate Prediction Interpretation and Evaluation Workshop .................................... 33and Tom Pagano

Douglas LeComte Climate Prediction: ENSO vs Non-ENSO Conditions ............................................. 29

Alex McCord Fire and Emergency Management in Arizona ............................................................ 55

Ron Melcher A Brief Introduction to Climate Forecast Resources for Fire Management ................ 55

Barbara Morehouse Climate and Human Impacts on Fire Regimes in Forests and Grasslandsand Steve Yool of the U.S. Southwest ............................................................................................... 41

Ron Neilson MAPSS: Mapped Atmosphere Plant Soil System....................................................... 20

Rick Ochoa NIFC Resource Impacts During the 2000 Fire Season ................................................ 8Weather and the 2000 Fire Season .............................................................................. 7

Jonathan Overpeck Climate, Fire, and the Need for a National Climate Service ...................................... 47

Tom Pagano and Climate Forecasts Overview for the Southwest .......................................................... 58Holly Hartmann

Roger Pulwarty Interagency Integrated Assessments ........................................................................... 40

Kelly Redmond Recent Developments in Data Access at Western Regional Climate Center ............... 15

John Snook The Role of Weather and Climate at Cerro Grande: A Quick Overview ................... 17

Tom Swetnam Climate and the 2000 Fire Season in the Southwest .................................................. 52

Anthony Westerling Climatology of Western Wildfire and Experimental Long-Range Forecasts ofWildfire Season Severity ............................................................................................ 25

Klaus Wolter Climate Forecasts for 2000 and 2001 .......................................................................... 9Climate Modeling Overview and Climate Forecast for 2001 ..................................... 22

3


Weather, Climate, Fire

The 2000 Fire Season and Beyond

Climate and Fire: Framing the Issues

in Context of the 2000 Fire Season

Tim Brown (Desert Research Institute)February 14, 2001 8:45 AM

La Niña and the 2000 Fire SeasonThe 2000 fire season was one of the most severe anddevastating fire seasons in recent decades. There were92,250 wildfire starts and 7,393,493 acres burned ac-cording to end of season National Interagency FireCenter (NIFC) statistics. On August 29, 2000, thepeak day of the fire season, 84 large fires were burning,including 1,642,579 acres burning in 16 states, andthe following resources were in use:

• 28,462 fire personnel• 667 crews• 1,249 engines• 226 helicopters• 42 air tankers

In order to put the key issues of this workshop intoperspective, I have selected some statements about the2000 fire season from NIFC. The statements refer, inlarge part, to NIFC’s interpretation of the effects of LaNiña on the weather and climate of the continentalU.S. during 2000. Accompanying illustrations showthat patterns of U.S. temperature, precipitation anddrought associated with extremes in La Niña are fairlypredictable. These patterns also show predictable spatialand temporal variation (Figure 1) during extreme years.

• “A pool of cool water in the Pacific Ocean deter-mined much of the country’s weather in the lasttwo years. La Niña changed normal weather pat-terns when it formed, and it’s still dominating theweather as it wanes.” NIFC Public Statement,August 2000

• “La Niña usually brings dry weather to the south-ern states and that is a big part of why Florida andthe Southwest have had such a severe fire season.La Niña has spread dry weather to the West thisspring and summer, and even though it is waning,the weather pattern is already set for the rest of the

summer and fall. Hotter and drier than normalweather is on tap through September.” NIFC Pub-lic Statement, August 2000

• “The Southwestern monsoon, which usually endsthe fire season in Arizona and New Mexico inearly July, has been sporadic this year. It’s into Au-gust, and the Southwest is still having an activefire season.” NIFC Public Statement, August 2000

• “So, here’s the summary: hot temperatures, lowrelative humidities, little or no precipitation,

Figure 1. Risk of extreme wet or dry years during La Niñafor spring (MAM, March-May; top) and summer (JJA,June-August; bottom). Note that the risk of extreme dryconditions migrates from the southern tier to the northerntier of the U.S. as the fire season progresses. Map source:NOAA-CIRES Climate Diagnostics Center.

4

plenty of wind, and the consequence is easy to pre-dict: the potential for a nasty fire season...None ofthis was unexpected by the federal firefighting agen-cies. They expected a tough season and planned ac-cordingly.” NIFC Public Statement, August 2000

• “As a result of La Niña and its influence onweather patterns, a combination of dry fuels anddry, hot weather led to what some are declaringone of the most severe wildland fire seasons inU.S. history.” NIFC Public Statement, October2000

The sequence of Standardized Precipitation Index(SPI) percentile of non-exceedance maps shown in Fig-ure 2 indicates that even though short-term conditionsin the spring of 2000 along the northern Rockies(around the Idaho-Montana border) were relativelymoist, long-term drought (including conditions 24months prior to summer 2000 in the NorthernRockies) produced conditions that led to extreme firedanger in that region during the 2000 fire season. Thepurpose of the SPI is to assign a single numeric value

to the precipitation that can be compared across re-gions with markedly different climates (McKee et al1993; 1995; Guttman 1998; 1999). The SPI was de-signed to explicitly express the fact that it is possible tosimultaneously experience wet conditions on one ormore time scales, and dry conditions at other timescales, often a difficult concept to convey in simpleterms to decision-makers. The SPI percentile of non-exceedance indicates the degree of “unusualness” of aparticular value of SPI. A value of 0 means that zeropercent of the other values in the record do not exceedthat value, or in other words, that all other values ex-ceed that value, i.e., the value in question is so low thatit seldom if ever occurs. A value of 50 indicates thathalf of the historical values are higher and 50 percentare lower. Values near 50 are not unusual; values near 0or 100 are very unusual. The SPI may have highervalue for fire management than the Palmer DroughtSeverity Index (PDSI) due in part to its ability to inte-grate precipitation occurrence over long time periods.

The maps of precipitation percentiles for summer2000 (Figure 2) make an important point, i.e., that it

Figure 2. Maps of Standardized Precipitation Index (SPI) percentile of non-exceedance for various lead times from thespring and summer of 2000. Five-month lead from the end of March 2000 (upper left); 6-month lead from the end ofAugust 2000 (upper right); 12-month lead from the end of August 2000 (lower left); 24-month lead from the end of August2000 (lower right). Map source: Western Regional Climate Center.

5


is important to look at long-term conditions, with re-gard to assessing the status of fuels and fire danger.Looking back one or two years shows that the South-west and, in particular, the northern Rocky Mountainstates exhibit precipitation amounts below the lowestone-third of the historical record back to 1895. In fact,Figure 3 shows that Montana climate division 2(Southwestern Montana) experienced below averageprecipitation conditions overall for 72 months prior tothe height of the 2000 fire season. Moreover, Montanadivision 2 continues to experience drought conditionsat present. However, if you only looked at conditionsfor the winter prior to March 2000, when some sea-sonal resource allocation decisions were being made,one would have seen a vastly different picture and per-haps drawn different conclusions, due to the recentwinter precipitation.

Some Uses of Climate Information During 2000SPI is just one of many climate variables relevant to afire season. The program for Climate, Ecosystems andFire Applications (CEFA) tracked the use of climateinformation by various collaborators in the fire man-agement community during the 2000 fire season. Wenoted that some of the most frequently used climateinformation included the following:

• Local and regional anomalies• Large-scale climate (La Niña)• Regional climate systems (southwest mon-

soon)

• Climate forecasts• Climate impacts and assessments

However, the mere use of information brings up ques-tions regarding the accessibility and usefulness of theproducts currently available. Such questions, impor-tant to the nature of this meeting, include the follow-ing:

• Is current information accessible?• Is current information sufficient?• Is climate information being satisfactorily in-

corporated into planning and decision-makingprocesses?

• How is climate information being used?• Is sufficient training and education in the use

of climate information available?• In what areas can the use of climate informa-

tion be improved?

CEFA gathered information from users through ourGreat Basin Remote Access Weather Station (RAWS)Network Analysis project. We found a wide array ofuses for these data including the following:

• Prediction or assessment of fire severity basedon historical information

• Examination of fire history• Fire investigations• Court cases• Post-fire erosion and erosion potential

Figure 3. Standardized Precipitation Index for Montana climate division 2 (Southwestern Montana)going back to 72 months prior to January 2001. The graph indicates long-term precipitation deficit forthis region. See text (above) for an explanation of the SPI. Graph source: Western Regional ClimateCenter.

6

• Historic season ending events• Risk appraisals for wildland fire use• Prescribed burn planning• Rehabilitation• Budget analysis• Fire behavior• Fire severity funding requests• Development of programmatic fire manage-

ment plans• Ground water and hydrologic assessments• Summaries to visitors and visitor guides• Wildlife impacts• Forest health assessment• Soils studies• Vegetation change and response studies

Using Climate ForecastsAreas of special interest with regard to use of climateinformation for fire management include the use oflong-term climate forecasts and the effect of climatevariability and climate change on fire regimes and firemanagement. The NOAA Climate Prediction Center(CPC), as well as several other agencies, produces cli-mate forecasts with long-lead times of up to one yearin advance. Unfortunately, it is relatively difficult fortypical users and decision-makers to find informationon how to interpret forecast maps and assessments offorecast skill. Several questions germane to the use of cli-mate forecasts by fire managers, include the following:

• How well is uncertainty understood and por-trayed?

• How are climate forecasts used in decisions?• What role do climate forecasts currently play

in planning?• How could climate forecasts be better utilized

in planning?• Do climate forecasts have sufficient skill for

use in planning?• Are climate forecasts easily accessible to users?

Climate variability and changeAs Figure 4 shows, there has been a long-term increasein global land-based and marine sea surface tempera-tures that has been especially pronounced during thepast two decades. Paleoclimate studies have shown pro-nounced natural climate variability on the scale of de-cades-to-centuries. Thus, it is incumbent upon us totake into account the effects that climate variabilityand possible human-induced climate change mighthave on fire regimes. The following questions come tomind:

• Can climate variability and change be adoptedinto longer-term management strategies?

• Can budgets and planning be invoked forlonger than a fiscal year?

• How will climate variability and change affectcomponents of fire danger, such as fire severityand fire potential?

• What are the human and environmental im-pacts of fire danger change and variability (eco-nomics; resources, wildland/urban interface)?

ReferencesGuttman, N.B., 1998. Comparing the PalmerDrought Index and the Standardized Precipitation In-dex. Journal of the American Water Resources Associa-tion, 34(1), 113-121.

Guttman, N.B., 1999. Accepting the StandardizedPrecipitation Index: A calculation algorithm. Journalof the American Water Resources Association, 35(2),311-322.

Jones, P.D., M. New, D.E. Parker, S. Martin, and I.G.Rigor, 1999. Surface air temperature and its changesover the past 150 years. Reviews of Geophysics, 37,173-199.

McKee, T.B., N.J. Doesken, and J. Kleist, 1993. Therelationship of drought frequency and duration of timescales. Eighth Conference on Applied Climatology,American Meteorological Society, Jan 17-23, 1993,Anaheim CA, pp. 179-186.

McKee, T.B., N.J. Doesken, and J. Kleist, 1995.Drought monitoring with multiple time scales. Ninth

Figure 4. Time series compiled jointly by the ClimaticResearch Unit and the UK Met. Office Hadley Centreshowing the combined global land and marine surfacetemperature record from 1856 to 2000 (Jones et al. 1999;http://www.cru.uea.ac.uk/cru/info/warming).

7


Conference on Applied Climatology, American Meteo-rological Society, Jan 15-20, 1995, Dallas TX, pp.233-236.

Related ResourcesProgram for Climate, Ecosystem and Fire Applicationshttp://www.dri.edu/Programs/CEFA/

Standardized Precipitation Indexhttp://www.wrcc.dri.edu/spi/spi.html

National Interagency Fire Centerhttp://www.nifc.gov/

Weather and the 2000 Fire Season

Rick Ochoa (National Interagency Fire Center)February 14, 2001 9:15 AM

The 2000 wildland fire season was one of the most se-vere in the last 75 years, not only in terms of the num-ber of acres burned, but also due to its duration andsize. The fire season began on January 1 with a smallfire in Florida and ended in late December in southernCalifornia (Figure 1). Highlights of the 2000 seasoninclude:

• 92,250 fires burned 7,393,493 acres• Over 860 structures lost• Estimated cost of fire suppression is $1.6 a

billion• Over 30,000 firefighters and support person-

nel involved

Precursors to the 2000 Fire SeasonWhile the fire season may have officially started with asmall fire on New Year’s Day, the seeds of the 2000 fireseason began with the emergence of La Niña in 1998.Florida experienced one of their worst fire seasons dur-ing the spring and summer of 1998. Hot and dry con-ditions during August-October, 1998 in the West leadto a very active fire season in Idaho and Montana.Even though the winter of 1998-99 saw record-break-ing snows in the Pacific Northwest, much of the areaeast of the Mississippi suffered under a severe droughtthat began in 1998. Fires were quite active in theFlorida during the spring of 1999 and also in the Ap-palachians during the fall. In 1999, the West had awindy spring, followed by a hot dry summer, includingdry lightning in August and September, and topped offby a warm and dry fall. This resulted in major fires for

California and over 1.7 million acres burned in Nevada.The 2000 wildland fire season was the culmination ofmany factors (Figure 2) including:

• Drier and warmer than normal weather since1998, due primarily to La Niña.

• Extremely dry fuels...at record dryness levelsin some areas.

• Greatly reduced spring rains, including failureof May-June rains in the Northwest andNorthern Rockies. Poor relative humidity re-covery and general moisture deficit.

• Dry lightning events.

Some of the hardest hit areas were:

• Idaho (1.36 million acres), including ClearCreek Complex (216,961 acres)

• Montana (0.95 million acres), including Val-ley Complex (292,070 acres)

Figure 1. Map of fire locations for the record 2000 fireyear (Source: National Interagency Fire Center).

Figure 2. Significant climatic anomalies during 2000(Source: NOAA Climate Prediction Center).

8

• Alaska (0.75 million acres).• New Mexico (0.52 million acres), including

Cerro Grande (47,650 acres).

The fire season in the Northwest, Northern Rockiesand Great Basin ended, for the most part, with the ar-rival of the rains in early September. Fire activitypicked up in the Southeast during October, with largefires reported in Virginia, North Carolina, South Caro-lina, Kentucky and Tennessee. Fire activity in Decemberwas generally confined to Florida and California.

Related ResourcesNational Interagency Fire Centerhttp://www.nifc.gov/

NIFC Resource Impacts During the

2000 Fire Season

Rick Ochoa (NIFC); assisted by Jan Hendrick (NIFC)February 14, 2001 10:00 AM

The National Interagency Fire Center (NIFC; http://www.nifc.gov) in Boise, Idaho is the national supportcenter for coordination of wildland firefighting andother incidents. Member organizations include the Bu-reau of Indian Affairs, Bureau of Land Management,Forest Service, Fish and Wildlife Service, National ParkService, National Weather Service, and the Office ofAircraft Services.

Orders for firefighting resources originate from the in-cident and are passed to the local dispatch center. Ifthe needed resource is not available in the local area,the request is sent to one of the eleven GeographicArea Coordination Centers (GACCs; Figure 1). Simi-larly, if the needed resource is not available within theGeographic Area, the GACC forwards the request tothe National Interagency Coordination Center(NICC). In turn, NICC will find the closest availableresource and coordinate its movement with the af-fected GACCs. It is important to remember thatNICC is a coordination unit and the local incident re-tains command and control authority.

When competition for resources arises, a Multi-AgencyCoordinating Group (MAC) may be formed at the re-gional or national level. The MACs set priorities, allo-

cate and/or reallocate resources, develop and recom-mend contingency action plans and issue coordinatedsituation reports.

Commensurate with the large number of fires andacres burned (Figure 2), the NICC processed a record46,245 resource requests with 32,362 for overhead po-sitions alone (note: overhead mobilization positions in-clude technical experts, GIS specialists, etc.). In addi-tion to civilian firefighters, the military provided sixbattalions (500 personnel per battalion) with over1,500 total personnel from Australia, Canada, Mexico,and New Zealand. On August 29, 2000, the peak dayof the 2000 fire season, the following resources werecommitted nationally:

• 28,462 people• 1,249 engines• 226 engines• 42 airtankers• 18 Type I teams• 84 large fires (100+ acres in timber or 300+

acres in grass/brush)

Adjustments to resource allocations are accomplishedthrough severity funding and pre-positioning/re-posi-tioning of existing forces. Examples of severity fundinginclude: additional funds to bring crews on earlier thannormal due to increased fire risk, or extending airtanker contracts for a longer than normal fire season.Such resource adjustments save money in the long runand they allow for substantial decreases in lag time forgetting resources to the incident. Climate forecasts arevery important in order to coordinate pre-positioning.

One example of re-positioning that occurred during

Figure 1. Map of 11 U.S. GACCs.

9


Figure 2. Wildfires and acres burned reported to NICC. Note the steady increase in fires approximately every 4 years.

Climate Forecasts for 2000 and 2001

Klaus Wolter (NOAA-CIRES Climate DiagnosticsCenter)February 14, 2001, 10:30 AM

Tools of the TradeIn talking about the climate forecasts it is important togive some background on the tools of the trade. Thereare statistical tools and modeling tools. Statistical toolsinclude the following:

• Optimal climatological normals (OCN). Thisgives an indication of the long-term trend.

• Canonical correlation analysis (CCA). Thisshows the spatial evolution of anomalies.

• Persistence (or anti-persistence). The tendencyfor conditions to remain the same, relative topast months.

• Analogs. Situations similar to the past.• Composites. Looking at the average of similar

events in the past as a guide to what mighthappen in the future.

Among the statistical tools, I most often use OCN andCCA. Anti-persistence has not been too successful re-

the 2001 fire season was when a very strong windevent was forecast for August 19 in Montana. Inpreparation for this event, the Northern Rockies MACrequested several strike teams of engines and crews forpublic safety and structure protection. The nationalMAC tasked the various geographic areas for these re-sources, even though these areas were experiencingwildfires. The greater risk of wildfires moving throughcommunities necessitated this mobilization of forces.The crews and engines were in place by August 19, butfortunately were not needed as the strong ridgetopwinds did not mix down into the valleys. This exampleshows an excellent use of long range forecast informa-tion to enhance coordination efforts.

Related ResourcesNational Interagency Fire Centerhttp://www.nifc.gov/

10

cent years. I will examine model-based forecasts in mytalk on climate prediction, tomorrow.

The Seasonal Cycle and Interannual RainfallVariabilityI have examined some SNOTEL (SNOwpackTELemetry) data for the Southwest and determinedthat during ENSO events variability at these moun-tains stations is in sync with variability at valley sta-tions. Note that there seems to be a monsoon dipolebetween Arizona/New Mexico and states to the north,Utah/northern Colorado/Southern Montana, duringJuly and August. If we look at climatology, charts ofaverage precipitation by season, note that above nor-mal spring precipitation in the Southwest might nothelp you too much in terms of fire hazard conditions.In the Southwest, if you do not get precipitation byApril 1st, then conditions will be dry during thespring; in contrast over the plains states and Montanathere is a significant component of spring precipita-tion. There some interesting patterns with regard tocoherence in signal between COOP (National WeatherService Cooperative Observational Network)/NCDCU.S. climate division data, which are mostly based atvalley locations, and SNOTEL data, which are basedat mountain locations. When these valley and moun-tain data are used independent of each other it doesnot give that good an idea of what is going on, al-though there are some core regions of spatial and tem-poral coherence. In Colorado, when looking at vari-ance in precipitation patterns in stations within theseregions I see little coordination between valley stations,e.g., Denver and many of the mountains stations (e.g.,Alamosa), which is very perplexing. In the Southwest,for example, western New Mexico does not have agood regional monsoon signal; most of the signal is lo-cal, at isolated stations; this is in contrast to the signalin central and southern Arizona. So I would say thatthere is a pressing need for more station data.

The State of ENSO — 3 Years of La Niña, and StillNot Dead!Knowing the state of ENSO has helped increase theskill of climate forecasts. At present, we’re in the thirdyear of La Niña conditions, (though these are relativelyweak La Niña conditions), and there is a possibilitythat these conditions will persist. It is important, how-ever, to note that we’ve never had four La Niña years ina row. Only the SST model is calling this year a LaNiña year. Thus, as we’re likely to transition out of LaNiña conditions, 2001 will be a tough year to forecast.During the last 50 years, 2000 was only the 8th stron-

gest La Niña (Figure 1). It was an unusual year, and LaNiña conditions started very late. If we continue tohave La Niña conditions in 2001 it will be verystrange. During a La Niña winter, there is three timesthe risk for a dry spring in the Southwest. Neverthe-less, with La Niña the strength of the event has less dowith the U.S. precipitation outcome, whereas duringEl Niño the greater the strength of the event, thehigher the amplitude of teleconnected outcomes.

CPC/IRI Forecasts For Water Year 2000 — WhatHappened?Reviewing summer 2000, the hallmark of 2000 was inthe temperature signal (Figure 2). Many temperaturerecords were broken, temperatures were very highthroughout the western U.S., there was enhanced firedanger, and especially high temperatures in the South-west (Figure 3). I will point out that we did not have agood forecast for climatic conditions in the Northwestand Intermountain West during 2000 (Figure 4).

CPC/IRI Forecasts For The Rest Of Water Year 2001With regard to 2001 forecast, as I said before, this willbe a tough year. We have some skill predicting between3-10 days for precipitation; however, beyond threedays, prediction remains difficult. If we look back atthe October 2000 climate forecasts and the January2001 forecast for February-April 2001, you will note achange. The October forecast indicates warm condi-tions throughout the West and an enhanced probabil-ity of wet conditions in Texas (based on trend). In con-trast, the January forecast indicates far less confidence

Figure 1. Multivariate ENSO Index (Wolter & Timlin, 1993;1998) for La Niña events. Note that the recent (1998-2001) La Niña is not the strongest; however, its behaviorhas been quite erratic. (http://www.cdc.noaa.gov/~kew/MEI/mei.html)

11


Figure 2. NOAA/CPC climate forecast for summer 2000. “CL” indicates that no prediction has been made. (http://www.cpc.noaa.gov/products/predictions/multi_season/13_seasonal_outlooks/color/seasonal_forecast.html)

Figure 3. U.S. climate division summer 2000 temperatureanomalies. Contrast these with the predicted shift inprobabilities of temperature in Figure 1. (http://www.cdc.noaa.gov/USclimate/USclimdivs.html)

Figure 4. U.S. climate division summer 2000 temperatureanomalies. (http://www.cdc.noaa.gov/USclimate/USclimdivs.html)

12

for temperatures above the 1961-1990 mean, andthere’s no indication of enhanced probability of wetconditions in Texas. The January forecast shows an en-hanced probability of below normal precipitation inFlorida, based on strengthening of La Niña conditionsduring the past two months. These contradicting indi-cators, and the lack of consensus on future La Niña-re-lated ocean temperatures (about which I will speak to-morrow), make for great uncertainty in the forecasts.

If we look at the skill of temperature predictions dur-ing recent winters, we find the following: predictionswere very good during the winters of 1997 and 1998,whereas predictions for fall 2000 were very bad. Since1999, precipitation predictions have generally beenpoor.

ReferencesWolter, K., and M.S. Timlin, 1993: MonitoringENSO in COADS with a seasonally adjusted principalcomponent index. Proc. of the 17th Climate Diagnos-tics Workshop, Norman, OK, NOAA/NMC/CAC,NSSL, Oklahoma Clim. Survey, CIMMS and theSchool of Meteor., Univ. of Oklahoma, 52-57.

Wolter, K., and M.S. Timlin, 1998: Measuring thestrength of ENSO - how does 1997/98 rank? Weather,53, 315-324.

Related ResourcesInternational Research Institute for Climate Prediction(IRI) Forecasts:http://iri.columbia.edu/climate/forecast/

NOAA-CIRES Climate Diagnostics Center (CDC)http://www.cdc.noaa.gov/

CDC El Niño and La Niña Pageshttp://www.cdc.noaa.gov/ENSO/

NOAA/NWS/CPC Suite of Official Forecastshttp://www.cpc.ncep.noaa.gov/products/predictions/

SNOTEL Datahttp://www.wrcc.dri.edu/snotel.html

Fire Forecasts for 2001

Tim Brown (Desert Research Institute)February 14, 2001 11:00 AM

The main message of this talk is that, reviewing cur-rent conditions and looking at model forecasts for thespring through midsummer, there is still a high firerisk for the West in 2001. Fire risk is particularly highfor the Pacific Northwest.

Review of Recent Precipitation and StreamflowConditionsWater year precipitation both at SNOTEL(SNOwpack TELemetry, an automated system to col-lect snowpack and related climatic data in the WesternUnited States) sites (Figure 1) and for NOAA climatedivisions is particularly low in the Pacific Northwest/Northern Rockies and near to above average in theSouthwest. As of February 1st, 2001, streamflow fore-casts suggest spring and summer streamflow at lessthan 70% of the 1961-1990 average for almost all ofthe western U.S. What is particularly significant aboutthis is that a large portion of the western U.S. receivesthe majority of its precipitation during the winter andearly spring.

Climate Forecasts for 2001A variety of consensus climate forecasts and ensembleclimate forecast model results suggest a greater likeli-hood of above average temperature and below averageprecipitation for the West this spring and summer. TheClimate Prediction Center (CPC) long-range seasonaltemperature outlooks show a high likelihood of above-average temperatures throughout the Southwest and

Figure 1. SNOTEL basin average water year (October-present) precipitation for the western U.S.

13


Figure 2. (a) NOAA Climate Prediction Center (CPC) seasonal precipitation outlooks for 2001-2002. Contours indicatepercent change in probability of A above-average and B below-average precipitation (based on the 1961-1990 average).CL indicates that no forecast has been made. (b) CPC seasonal temperature outlooks for 2001-2002 (contours andsymbols are the same as in (a), except for temperature).

A

B

14

coastal West for the spring and throughout the fire sea-son (Figure 2a); CPC long-range seasonal precipitationoutlooks show a slightly greater likelihood of above-av-erage spring precipitation in the Pacific Northwest anda slightly greater likelihood of below average of fire sea-son precipitation in northern California and Nevada(Figure 2b). Remember, that the climatological meanprecipitation for the Pacific Northwest is fairly low forthe spring; therefore, increased spring precipitationwill not likely make up for the lack of winter precipita-tion in the Pacific Northwest. The International Re-search Institute for Climate Prediction (IRI) predicts agreater likelihood of above-average temperaturesthroughout the West for April-June 2001, with par-ticularly high likelihood of above-average temperaturesin the Pacific Northwest and the Southwest. A varietyof ensemble forecasts show similar results. For ex-ample, the NCEP (National Centers for Environmen-tal Prediction), ECHAM (European CommunityHamburg), NCAR CCM3 (National Center for At-mospheric Research Community Climate Model) en-semble forecasts all indicate below average spring andearly summer precipitation, especially in Californiaand the central and southern Rockies.

On the basis of these analyses, I predict high fire riskin West. Of course, fires are both lightning and humancaused, and lightning predictions would be helpful forfire forecasting, but about half of all fires are humancaused and climate predictions probably give some in-dication of the risk of spread.

The GSM, a weekly forecast model produced by the

Experimental Climate Prediction Center(ECPC) at Scripps (running a variationon the NCEP model) indicates contin-ued dry conditions throughout theWest, with some precipitation in latespring in the Intermountain West. GSMpredictions also include forecasts of FireWeather Index (FWI), wherein tempera-ture humidity and wind are used to cre-ate an integrated measure of fire danger.Such integrated measures have not beenevaluated thoroughly in the “fire busi-ness.” The GSM FWI forecast for spring(Figure 3) suggests elevated fire dangerin the Pacific Northwest and northernCalifornia, as well as across the centraland southern Rockies and the Plainsstates.

SummaryIn summary, fire danger is still pretty high for most ofthe western U.S., especially for the Pacific Northwestand northern California, where winter conditions havebeen exceptionally dry. Again, increased precipitationin the Pacific Northwest this spring will not be able toameliorate precipitation deficits that have accumulatedover the winter months. However, it is important toevaluate separately spring precipitation predictions forareas such as the central Rockies, where spring is theseason of highest precipitation. The most importantthings about fire prediction include fuel conditions,weather/climate and topography. CEFA in conjunctionwith researchers at Scripps ECPC will be incorporatingmeasures of these variables in order to produce weeklypredictions of fire danger.

Related ResourcesProgram for Climate, Ecosystem and Fire Applicationshttp://www.dri.edu/Programs/CEFA/

SNOTEL Data (Western Regional Climate Center)http://www.wrcc.dri.edu/snotel.html

Scripps Institution of Oceanography ExperimentalClimate Prediction Centerhttp://ecpc.ucsd.edu/ecpc.html

Figure 3. GSM (Global Spectral Model) Fire Weather Index forecast forMarch-April, 2001, produced by the Experimental Climate PredictionCenter (ECPC) at Scripps Institution of Oceanography.

15


Recent Developments in Data Access

at Western Regional Climate Center

Kelly Redmond (Desert Research Institute, WesternRegional Climate Center)February 14, 2001 1:15 PM

The Climate of the 2000 Fire Season and CurrentConditions.As a starting point, we begin by looking at the averagenumber of acres burned per decade in the U.S. Farmore acres burned per year before 1940 than duringthe past several decades; in fact, during the 1930s moreacres burned than the total for the last few decades.Confining attention to the changed regime of the past40 years, the active 2000 fire year emerges as compa-rable to three or four of the years since 1960 (e.g.,1996, 1988, 1969, 1963). For the recent winter, weobtain a historical perspective from Standardized Pre-cipitation Index percentiles for 1-4 month periods ex-tending back from January 2001 through the prior 1to 4 months. The heart of winter was quite dry, butthe extremely unusual wet conditions in the Southwestduring October make it appear that this entire periodwas moist. Every winter brings a mystery, it seems. LaNiña is present for a third consecutive winter. Typi-cally, La Niña brings dry winters to the Southwest andwet winters to the Northwest. However, Southwestprecipitation has been near to slightly below averagehis winter, and the Northwest is experiencing recorddrought, almost the exact opposite of widespread ex-pectations. So, why?

Right now (February 6, 2001) moderatedrought conditions are prevalent over thePacific Northwest and northern Rockies(Figure 1). Snowpack is far lower thanusual, and there has been very low precipi-tation in the Columbia River basin thisyear. A quick look at precipitation climatol-ogy for the first half of the water year (Fig-ure 2a) and for the spring season (Figure2b) shows us that there’s a very low chanceof recovery from current drought condi-tions in the Pacific Northwest. This is be-cause normally about 80% of annual pre-cipitation in the Pacific Northwest is re-ceived during the months of October-March, whereas only a small percentage isreceived between April-June. The situationhas been a little bit different for the north-ern Rockies, where a climatological second

maximum of spring precipitation allows for some de-gree of recovery before summer sets in.

The situation is quite different for the southwesternU.S. The Southwest receives upwards of 40% of an-nual precipitation during the summer monsoon (Fig-ure 3), which usually starts around the 4th of July. Anew research initiative called the North AmericanMonsoon Experiment (NAME) is forming, in order to(1) understand the monsoon (summer rain) and whatdrives it and (2) examine ways in which this experi-ment could produce improved information about themonsoon that would be useful for decision-making.The area affected encompasses New Mexico and Ari-zona, and northern Mexico.

Climate MonitoringThe Western Regional Climate Center (WRCC)monitors several major western data sources to trackclimate in the region. WRCC is the repository for animportant network used extensively by the fire com-munity known as RAWS (Remote Access Weather Sta-tion). This network currently consists of 930 stations(1700-1800 hundred have existed over time). RAWShas primarily a summer orientation, and data arrivehourly via GOES satellite. WRCC is in the process ofdeveloping web tools to improve the access to RAWSinformation. By way of example, we plan to developmap-clickable access to the following types of prod-ucts:

• daily summary• monthly summary

Figure 1. U.S. Drought Monitor (http://enso.unl.edu/monitor/monitor.html) for February 6, 2001 indicates drought conditionsthroughout the Pacific Northwest and Northern Rockies, due to lowprecipitation and extremely low winter snowpack (see text).

16

• graph of the last seven days• historical time series (since 1985)• wind statistics and wind roses

The basic time increment of the raw data is hourly.Among the elements which are measured are:

• wind speed and direction• air temperature and fuel temperature• humidity• solar radiation (about 60 percent of the sites).

In addition we would like to make available compositedaily summaries for every station. It is important tonote that RAWS stations are not all-season gauges;they are solar powered, and thus do not have sufficientheating to melt solid winter precipitation. The bestspatial coverage for RAWS data stations is between5000 and 7000 feet in elevation. Given the region’s ex-treme spatial heterogeneity, RAWS does not currentlysample all major western vegetation zones in all geo-graphic areas.

However, there are excellent complements to theRAWS network that greatly extend climate andweather station coverage of the western states (andAlaska/Hawaii). The National Weather Service oper-ates about 2500 cooperative observation stations, andat higher elevations there is the SNOTEL (SNOwpackTELemetry) network, with about 700 sites run byUSDA/NRCS. Each of these networks measures differ-ent quantities in support of differing missions. Hope-fully new RAWS stations can be sited to provide long-term climate data for currently insufficiently sampledelevation bands or ecological zones.

Climate Information Needs For Fire ManagementFor sound natural resources management, we need toprovide climate information that can be interpretedfrom a long-term perspective (several decades). Needsby the fire community for atmospheric data range overtime scales from hours to many decades, and theRAWS network has both meteorological and climato-logical characteristics. The condition of range and veg-etation with respect to fire susceptibility is affected byevents over the past several seasons, and often longer.For this reason, there is strong incentive to improve theRAWS network to be able to provide accurate andtimely data all year long, including winter. We need tobe working toward synergistic operation and coordina-tion of all federal networks, to obtain the most valuefor the public dollar, to provide requisite information

Figure 3. Percent of annual average precipitation (1961-1990) for July-August. Joint product of the WesternRegional Climate Center and OSU/PRISM.

Figure 2. Percent of annual average precipitation (1961-1990) that falls in (a) October-March and (b) April-June.Joint product of Western Regional Climate Center andOSU/PRISM.

17


Pre

cipi

tatio

n (in

ches

)

The Role of Weather and Climate at

Cerro Grande: A Quick Overview

John Snook (United States Forest Service)February 14, 2001 1:45 PM

Pre-Burn Background InformationBy way of introduction, the initial prescribed burn atUpper Frijoles, a higher elevation site, was very tamecompared to the raging Cerro Grande wildfire of threeto seven days later. Due to 0.22 inches of rain that oc-curred April 29 to May 1, the initial burn would barelycarry in the timber fuels on evening of May 4. In con-trast, the wildfire raged uncontrollably a few days later,following significant changes in wind, fuel type, as-pect, and in an adjacent, low elevation, environmentwhere drought-effects were brought to the forefront.

A graph of precipitation at stations within the regionshows that all stations recorded lower than normalwinter precipitation (Figure 1). On May 6, droughtconditions were worse in southern and western NewMexico than in Bandelier (Figure 2). Nevertheless,conditions at Bandelier were still dry. In fact, the localski area never opened that winter.

Just days before the fire, Jeff Baares presented a studyon wind and fire danger at Los Alamos National Labo-ratory. The study looked at the joint probability of theco-occurrence of strong winds, from South to WNW,and high or very high fire danger. The data used cov-ered the period April-June for 1980-1998. For the pur-pose of the study, fire danger was assessed using theBandelier, NM energy release component. Strong

at the spatial scales needed, and to develop the longdata sets which will be expected by future generationsin order to perform the more sophisticated analysisthey will require. In addition, there is an equally strongneed for web-based tools to efficiently access and sum-marize this wealth of data that exist.

Related ResourcesDesert Research Institute, Western Regional ClimateCenter http://www.wrcc.dri.edu/

North American Monsoon Experiment (NAME)http://www.cpc.ncep.noaa.gov/products/precip/mon-soon/NAME.html

winds were defined as 15-minute winds greater than10mph. Winds this strong typically produce instanta-neous gusts of approximately 30-40 mph.

FindingsThe conclusions of Baares’ study were as follows:

• March-June is the windiest time of year in NewMexico.

• The aforementioned combination of fire danger,wind direction, and wind speed occurs over a 3-day period once every four years. When such 3-day periods occur, there is usually more than oneepisode in that year.

• Thus, a major fire moving up to the edge of theLos Alamos National Laboratory is not only cred-ible, but likely, with a frequency of one occurrenceper 10 years.

The following discussion documents the conditionsleading up to the burn and the escape of the burn. Thegoverning weather patterns prior to the Upper Frijolesburn were as follows:

• Most of April 2000 was warmer and drier thannormal, under the influence of strong high pres-sure. April 29 to May 1 was an exception, with aweather system bringing 0.22 inches of precipita-tion to the site. The high pressure ridge re-intensi-fied during the first four days of May and re-mained strong from May 4-7

• The weather on the day of the Upper Frijolesburn, Thursday May 4, 2000, recorded by the

Figure 1. Comparison of 1999-2000 winter with normalprecipitation at Los Alamos area stations.

18

Figure 2. Drought conditions prior to the Cerro Grande burn.

19


portable weather station (9170 ft.):

• Sky: Sunny skies• Maximum Temperature: 72°F• Minimum Relative Humidity (RH): ~15-20%• Minimum Temperature: 48°F

A test burn was conducted at about 7:20 p.m. Theweather conditions at that time were as follows:

• Upper Elevation Temperatures: in the 50s• RH: ~25-30%• Evening Winds:• On the ridgetops: NW 8-12 mph• On the slopes:1-5 mph

Thus it was still not very windy during the test burn,but conditions started to change on Friday and Satur-day in that winds increased a bit. The weather duringthe next several days, Friday, May 5 to Monday May 8,was as follows:

• Friday May 5: Mostly sunny through midday,some clouds p.m. Warm temperatures, with mini-mum RH 14-18%; 20' winds W to SW, increasingto 15-18 with gusts of 20-22 mph.

• Saturday May 6: Cloudier and a little cooler, withRH similar to Friday. Eye level winds were SW toWest 1-5 mph with gusts of 8-11 mph.

• Sunday May 7: Variable clouds and little tempera-ture change; RH rose to 20-30%; 20' winds in-creased to SW 10-15 with gusts of 28-40 mph. Atthis point the fire escapes along the south perim-eter. There were attempts to contain the escape,which led to a bigger problem.

• Monday May 8: Winds again fairly strong andgusty. The passage of a low-pressure trough shiftedthe high elevation winds from SSW to W or NWat 3:00 p.m.

Within the prescribed fire area, small and medium fu-els on the ground did not burn, due to local groundmoisture. This is because smaller dead fuels react morequickly to daily weather changes. The long-termdrought’s biggest effects were on live fuels and largedead fuels. These were more affected by seasonal trendsthan by daily weather.

LessonsShorter time-frame weather events, such as the lateApril rainfall in northern New Mexico, can tempo-rarily and/or locally mask the criticality of the “big pic-ture,” i.e., the long-term regional drought. This wasthe situation with the initial prescribed burn, and thefuel moisture conditions at the high elevation burnsite. Therefore, we need to look at all scales and condi-tions, both local conditions and wider regional condi-tions. One should not automatically conclude thatsimply because there is a drought in effect, every firewill have extreme fire behavior. Each fire situationshould be addressed on a case-by-case basis, and bothshort-term and long-term factors must be considered.

Related ResourcesBandelier National Monument Cerro Grande FireWeb Sitehttp://www.fs.fed.us/r3/sfe/fire/cerrogrande/

Bandelier National Monument Cerro Grande Pre-scribed Fire Investigation Reporthttp://www.nps.gov/cerrogrande/

Cerro Grande Fire Board of Inquiry Final Report andother Reportshttp://www.fire.nps.gov/fireinfo/cerrogrande/reports.htm

yrotarobaLlanoitaNsomalAsoLtastsuGdniWmumixaM0002,8-4yaM

etaD 4yaM 5yaM 6yaM 7yaM 8yaM

stsuG)hpm;noitcerid(

03EN 73W 03W 53WSW 04WS

emiT MP21:2 MP52:01 MA51:1 MP75:1 MP73:2

20

Beyond the 2000 Fire Season

MAPSS: Mapped Atmosphere Plant

Soil System

Ron Neilson (USDA Forest Service, PacificNorthwest Research Station)February 14, 2001

(Editor’s Note: Ron Neilson generously offered to givean impromptu talk on his innovative modeling of in-teractions between climate, vegetation and fire).

MAPSS is a vegetation distribution model that was de-veloped to simulate the potential biosphere impactsand biosphere-atmosphere feedbacks from climaticchange. MAPSS was originally a steady-state biogeog-raphy model, able to simulate a map of potential natu-ral vegetation under a long-term average climate.Emerging technology couples the biogeographical rulebase of MAPSS with two different ecosystem nutrientcycling models and a process-based fire model in orderto simulate the spatially explicit dynamics of vegeta-tion at landscape to global scales under both stable andchanging climates. These new dynamic vegetationmodels (DVM) will be useful for exploring manage-ment options at all scales from landscape to regional,national and global.

The climate component of the new models incorpo-rates data generated from PRISM (Parameter-elevationRegressions on Independent Slopes Model; http://www.ocs.orst.edu/prism/prism_new.html). PRISMuses point data, a digital elevation model (DEM), andother spatial data sets to generate estimates of climaticelements that are gridded and GIS-compatible.PRISM employs a coordinated set of rules, decisions,and calculations, designed to accommodate the deci-sion-making process an expert climatologist would in-voke when creating a climate map. The strong varia-tion of climate with elevation is the main premise un-derlying the model formulation. PRISM adopts the as-sumption that for a localized region, elevation is themost important factor in the distribution of tempera-ture and precipitation.

The conceptual framework for the MAPSS model isthat vegetation distributions are, in general, con-strained either by the availability of water in relation to

evapotranspirational demands, or the availability of en-ergy for growth. In temperate latitudes, water is theprimary constraint, while at high latitudes energy isthe primary constraint. The model calculates the leafarea index (LAI) of both woody and grass life forms incompetition for both light and water, while maintain-ing a site water balance consistent with observed run-off. Water in the surface soil layer is apportioned to thetwo life forms in relation to their relative LAIs and sto-matal conductance, i.e. canopy conductance, whilewoody vegetation alone has access to deeper soil water.

Biomes are not explicitly simulated in MAPSS; rather,the model simulates the distribution of vegetationlifeforms (trees, shrubs, grass), the dominant leaf form(broadleaf, needleleaf ), leaf phenology (evergreen, de-ciduous), thermal tolerances and vegetation density(LAI). These characteristics are then combined into avegetation classification consistent with the biomelevel.

The biogeochemistry component of the new DVMssimulates monthly carbon and nutrient dynamics for agiven ecosystem. Above- and below-ground processesare modeled in detail, and include plant production,soil organic matter decomposition, and water and nu-trient cycling.

The fire component simulates the occurrence, behaviorand effects of severe fire. Allometric equations, keyedto the lifeform composition are used to convert above-ground biomass to fuel classes. Fire effects (i.e., plantmortality and live and dead biomass consumption) areestimated as a function of simulated fire behavior (i.e.,fire spread and fire line intensity) and vegetation struc-ture. Fire effects feed back to the biogeochemistrymodule to adjust levels of various carbon and nutrientpools.

The vegetation component integrates long-term veg-etation and soil moisture responses to weather and cli-mate. Various climate scenarios can be used to projectvegetation change. The model simulates “natural” pre-European ecosystems (it does not yet include historicalor current land uses) and uses observed climate condi-tions to generate fire scenarios (i.e., biomass con-sumed).

21


We believe that MAPSS-based dynamic vegetationmodels, incorporating fire simulations, can be used todetermine which climate variables and scenarios gener-ate hot spots for future fire risk. We plan to use a fine-resolution model to evaluate this. In the future, we willbe incorporating land use and change to refine themodel.

Related ResourcesMAPSS (Mapped Atmosphere-Plant-Soil System)http://www.fs.fed.us/pnw/corvallis/mdr/mapss/

An Overview of the Joint Fire Science

Program and RFPs for 2001

Bob Clark (Joint Fire Science Program, NIFC)February 14, 2001

(Editor’s Note: Bob Clark generously offered to give animpromptu talk about the Joint Fire Science Programand upcoming research opportunities offered by theprogram).

The Joint Fire Science Program (JFSP), a six agencypartnership to address wildland fuels issues, was autho-rized and funded by Congress in October, 1997. Thesix agencies, designated by the Congress, are theUSDA Forest Service and five bureaus of the Depart-ment of the Interior: Bureau of Indian Affairs, Bureauof Land Management, National Park Service, U.S. Fishand Wildlife Service, and the U.S. Geological Survey.The purpose of the program is to provide wildland fireand fuels information and tools to specialists and man-agers who make wildland fuels management decisions.The information and tools will also help agencies de-velop sound, scientifically based land use and activityplans.

A Joint Fire Science Plan was prepared at the requestof the Congress. The plan describes four principal pur-poses. Task statements in requests for proposals (RFPs)are developed to further one or more of the principalpurposes. The four purposes are:

• Fuels inventory and mapping• Evaluation of fuels treatments• Scheduling of fuels treatments• Monitoring and evaluation of fuels

treatments

RFPs that are designed to answer specific questions orsolve specific problems related to wildland fuels issuesare issued periodically as funding is available (http://www.nifc.gov/joint_fire_sci/RFPs.htm). This year’sRFPs include some new elements, including interac-tions between flora, fauna, fuels and climate, as well asissues regarding water and cultural resources. The basicelements of two RFPs germane to this meeting are:

• Development of methods and systems to in-corporate weather and climate data into tacti-cal and strategic fire planning (rfp 2001-1)

• Development of decision support tools for fu-els and fire management (RFP 2001-2).

The JFSP requires face-to-face technology transfer tohand off research results to end-users. Moreover, allproposals must have a federal cooperator.

22

Climate Prediction

Climate Modeling Overview and

Climate Forecast for 2001

Klaus Wolter (NOAA-CIRES ClimateDiagnostics Center)February 15, 2001, 8:00 AM

Background to Forecasting and General CirculationModelsForecasting a complex system, such as the climate sys-tem, is a difficult task. This is because there is an enor-mous amount of chaos inherent in the climate system.If we use a pinball machine as an analogy, even giventhe same initial conditions, there are a myriad of pos-sible outcomes, although some will occur more oftenthan others. In order to determine the most likely out-come of the chaotic climate system, we run ensemblesof general circulation models and sea surface tempera-ture (SST) prediction models and extract the meanmodel prediction. General circulation models are runevery week, and monthly and seasonal predictions aredisplayed as an “anomaly plume.”

Recent Model PredictionsIf we look at recent SST predictions, for regions of thetropical Pacific that are sensitive ENSO indicators, theECMWF model shows a change over to El Niño con-ditions by spring, strengthening into summer; twoforecasts from the National Centers for EnvironmentalPrediction (NCEP) model also show increases in SSTsin the Niño 3.4 regions developing by late spring andstabilizing by summer. On the other hand, NOAAforecasts show conditions returning to normal by sum-mer and maintaining throughout the end of the year,whereas the University of Maryland model actuallyshows strengthening of La Niña conditions by summer2001. Thus, there is no definite consensus amongmodels, and no clear forecast. As I mentioned in myprevious talk, at this point in time, the indicators arenot clear for good predictions of summer precipitation.

The Most Recent Model ForecastsThe most recent seasonal precipitation forecast (fromFebruary 2001; Figure 1) from NCEP shows lowerthan normal precipitation across a wide swath of theUnited States. The strongest negative anomalies arecentered around Colorado and Northern California, as

well as particularly strong negative anomalies acrossFlorida. The dryness in Florida is probably a functionof La Niña conditions. Note that there is the slightpossibility that the Pacific Northwest may recover fromthe dry conditions during the winter.

The model predicts that these conditions will persistthroughout the spring and into the summer, taperingoff slightly during the summer. Temperature predic-tions indicate above-normal temperatures for theSouthwest and the Central Plains throughout thespring, tapering off during the summer.

Recent, as yet unpublished, model sensitivity studiesdone by Martin Hoerling of NOAA’s Climate Diag-nostic Center show that the location of tropical Pacificwarming during El Niño events is a strong influenceon winter precipitation anomalies across the UnitedStates. Preliminary results show that tropical SSTanomalies centered further west in the Pacific result ina more typical Northwest/Southwest dipole in precipi-tation, whereas those centered further east result in avery different pattern.

Recent Rainfall AnomaliesIn order to put the forecasts into perspective, we canlook at recent precipitation anomalies. We can see nor-mal to above-normal precipitation across the western

Figure 1. NCEP precipitation forecast (issued February2001) for March-May 2001. Units are mm/month anomalyfrom 1961-1990 mean. (http://www.emc.ncep.noaa.gov/research/cmb/atm_forecast)

23


Figure 3. U.S. climate division precipitation anomalies forOctober 2000-January 2001. Note the extremely dryconditions across the coastal states and PacificNorthwest. (http://www.cdc.noaa.gov/USclimate/USclimdivs.html)

Figure 2. U.S. climate division precipitation anomalies forOctober 2000, an unusually wet month in the Southwest.(http://www.cdc.noaa.gov/USclimate/USclimdivs.html)

Figure 4. U.S. climate division summer precipitationanomalies for the 14 lowest NAO values in the 20th

century. (http://www.cdc.noaa.gov/USclimate/USclimdivs.html)

U.S. (with the exception of western Washington state;Figure 2) during the fall of 2000, progressing to ex-tremely dry conditions across the coastal states andNorthern Rockies, with above-average precipitationover much of the Southwest and Central Plains (Figure3). If we refer back to Hoerling’s studies, we see thatwarm anomalies over the Central and eastern Pacificresult in a wet west coast, as happened during the1983 and 1998 El Niños. It is quite possible that,similarly, the location of La Niña anomalies has an ef-fect on U.S. precipitation anomalies, such as duringthis winter.

If we look forward to the summer season, if tropicalPacific SST anomalies, in fact, are positive by spring,then we can expect good Southwest monsoon condi-tions and dry conditions in the Midwest. However,our ability to predict summer monsoon rainfall has notbeen particularly good. We are finding that it is impor-tant to consider the Atlantic side of the continent, be-cause SSTs in the Caribbean also influence the mon-soon. An interesting angle is to look at the North At-lantic Oscillation (NAO), which has been mainlynegative this year. If we use historical negative NAOconditions as an analog for what might happen duringthe summer, we see a pattern of dry conditions in thePacific Northwest and wet conditions in Arizona (Fig-ure 4). However, we have not yet figured out all ofpieces of the puzzle. Certainly there’s a lot of work yetto be done on the empirical side, in order to figure outpertinent relationships.

ReferencesWolter, K., and M.S. Timlin, 1998: Measuring thestrength of ENSO - how does 1997/98 rank? Weather,53, 315-324.

Related ResourcesClimate Diagnostics Center (NOAA-CIRES)http://www.cdc.noaa.gov

ECMWF European Centre for Medium-RangeWeather Forecasts http://www.ecmwf.int/NCEP National Centers for Environmental Predictionhttp://www.ncep.noaa.gov

NAO The North Atlantic Oscillationhttp://www.ldeo.columbia.edu/~visbeck/nao/presenta-tion/html/NAO.htm

NOAA Climate Prediction Center (CPC) http://www.cpc.ncep.noaa.gov

24

Seasonal Climate Forecasts: Wildfire

Applications

Dan Cayan (Scripps Institution of Oceanography)February 15, 2001 8:20 AM

Seasonal Forecast CharacteristicsIn order to best understand seasonal climate forecasts,it is important to understand the characteristics thatdetermine the results of model runs. In general the fol-lowing characteristics of seasonal climate forecasts areimportant:

• boundary conditions, not model initializationconditions dictate model results

• models are an aggregate overtime and do notyield synoptic details

• most models have coarse spatial resolution• model results are probabilistic, not determinis-

tic ensembles• methods include: statistical, dynamical, statis-

tical/dynamical hybrid• climate forecast model skill has been poor-to-

modest• skill may depend upon any of the following

factors: season, state of ENSO, model• minimum skill is at 2-4 weeks

Thus far, seasonal climate forecasting does not accountwell for multi-year trends, a point that I will examineat length, below.

The International Research Institute for Climate Pre-diction (IRI) produces skill evaluations for 3 differentmodels, NCAR CCM (National Center for Atmo-spheric Research Community Climate Model), NCEP(National Centers for Environmental Prediction), andECHAM (European Community Hamburg). If welook, for example, at a skill map for the ECHAMmodel winter precipitation (January-March), correla-tions are mostly positive, but weak (especially forNorth America), and overall skill is pretty good (Figure1). For summer, correlations are far lower and the spa-tial pattern seems to be random. Again, model skill issubject to initialization conditions, spatial resolutionand, most clearly, season.

Some new studies by Sasha Gershunov of Scripps com-pares statistical and dynamical seasonal climate forecastmodels. If we look at Gershunov’s model validation forwinter 1998, we can see that the statistical model un-derestimates the total amount, because it cannot repro-duce the extremes; the RSM (dynamical) model over-estimates precipitation amount, but gets precipitationintensity pretty well. A hybrid of statistical and dy-namical models is considerably more skillful than theindividual approaches. Similarly, a hybrid forecastmodel has higher El Niño forecast skill.

Long-Term TrendsThe state of the ENSO system has a clear signal inlarge precipitation and large streamflow events, espe-cially during La Niña years, when stations in the Pa-cific Northwest exhibit a high number of largestreamflow events between January and July. However,skill in predicting precipitation and related streamflowpeaks, is mitigated by long-term trends. For example,since 1950 there has been an increase in spring(March-May) temperature, in excess of 2°C over thewestern U.S. and into Alaska and Canada. This trendhas been verified in phenological data on the first dateof lilac blooms, which now bloom about one week ear-lier than in the early historical record. Our data fromspring temperature trends in the Pacific Northwest, li-lac blooms and the date of spring snowmelt shows aclear positive relationship between the three, such thatduring the past 20 or so years both the date of 1stsnowmelt and the date of lilac blooms have occurredabout one week earlier than they did to prior to 1976(Figure 2).

Figure 1. Skill map for ECHAM (European CommunityHamburg) winter (January-March) precipitation forecasts,expressed as correlations between forecast and actualprecipitation. Source: IRI.

25


This research has powerful implications not just forthe hydrological cycle, but climate prediction for wild-fire, because both depend on the timing of snowmelt.Work by Mike Dettinger shows that there is a nonlin-ear trend in predicted and observed streamflow.Streamflow predictions from ensemble models arepretty accurate up until the point of snowmelt, thenensemble predictions vary greatly, probably due totrends in spring temperature. These regional trendspose a great challenge for medium-range climate pre-diction.

Wildfire ApplicationsTony Westerling of Scripps has produced some statisti-cally-based fire predictions, using Palmer Drought Se-verity Index (PDSI) as a predictor. Westerling’s re-search shows strong regional spatial coherence in firestarts. Correlations of seasonal fire activity, in variousparts of the country, with lagged PDSI show that, de-pending on the region, there are strong positive corre-lations between PDSI months in advance of the cur-rent fire season and seasonal fire activity. These laggedrelationships are related to fuel buildup and groundmoisture, so high PDSI values (high moisture) lead tofine fuel buildup. The fine fuels dry out as the PDSIdrops to drought levels during the current fire year.

Westerling has found that when wildfire measures(e.g., acres burned) are correlated with atmosphere cir-culation variables for the winter and spring prior to thefire season, as well as the summer of the fire season,large-scale atmospheric circulation and global-scale cli-mate anomalies become evident. For most of the West,wildfire activity appears not to be well related to ElNiño and La Niña, but rather to circulation patternsthat create first wet, then hot and dry conditions.Thus, there is an ability to predict fire danger, seasonsin advance. We do not specifically need El Niño/LaNiña conditions in order to predict fire activity. How-ever short-term events and ignition events are stillproblematic for seasonal prediction; lightning, SantaAna winds and hot spells are difficult to predict withany accuracy.

Needs for Future ResearchIn order to improve climate forecasts for wildfire appli-cations and statistical wildfire prediction models, weneed the following:

• more and better wildfire data• models to elucidate climate/weather/fire links• historical forecast archives

Figure 2. Pacific Northwest temperatures (°C), firstprinciple component of the date of western U.S. lilacblooms, first principle component of the date of westernU.S. first snowmelt, 1950-1998. Note the post-1976 shifttoward higher temperatures, earlier lilac blooms andearlier spring snowmelt. Source: Dan Cayan, ScrippsInstitution of Oceanography.

At Scripps, we are in the process of building an archiveof seasonal and medium-range forecasts back to 10years, with the cooperation of the Climate DiagnosticsCenter. Moreover we’re working on unraveling thelinks between climate and fire in the western U.S.

Related ResourcesScripps Institution of Oceanography Climate ResearchDivision http://meteora.ucsd.edu/

California Applications Programhttp://meteora.ucsd.edu/cap/

Climatology of Western Wildfire and

Experimental Long-Range Forecasts of

Wildfire Season Severity

Anthony Westerling (Scripps Institution ofOceanography)February 15, 2001 8:40 AM

The number and extent of wildfires in the westernUnited States each season are driven by natural factorssuch as fuel availability, temperature, precipitation,wind and relative humidity anomalies, and the loca-tion of lightning strikes, as well as anthropogenic fac-tors. It is well known that climate fluctuations signifi-cantly affect these natural factors at a variety of tempo-ral and spatial scales. It is of great interest to know howstrongly climate patterns affect wildfire, and further-more, whether there may be time lags imposed by cli-

26

mate forcing such that prediction skill may be possibleat a season’s or longer lead time. At longer lead times,fuel availability, being substantially determined byprevious seasons’ or years’ climate, may be amenableto even longer-range forecasting. A group of us atScripps have developed a seasonal fire severity fore-cast based upon a number of parameters, includingclimate factors.

DataWe compiled and combined some 330,000 individualfire reports from United States Forest Service (USFS),Bureau of Land Management (BLM) and the NavajoIndian Reservation (BIA) to construct a data set ofmonthly fire start counts and acres burned as 1° x 1°grid cells (Westerling et al., 2001a). For our analysis,we assumed that there is a dominant fuel type in eachgrid cell that can be classified as either heavy or fine forgeneral descriptive purposes. In the resulting series, lessthan 1% of the USFS data and less than 3% of theBLM data are excluded for lack of proper location in-formation or dates. Despite their limitations, an amal-gamation of these data sets yields a spatial and tempo-ral history that is of sufficient quality, spatial resolutionand duration to resolve regional characteristics of thewildfire season.

Climate data are represented by the Palmer DroughtSeverity Index (PDSI). The PDSI is an auto-regressivemeasure of combined precipitation, evapotranspirationand soil moisture conditions. It represents accumu-lated precipitation anomalies and, to a very small ex-tent, temperature anomalies. When the PDSI is nega-tive (positive), soil moisture is below (above) averagefor a location. PDSI has been used before in studies ofclimate and fire relationships and, despite the consider-able limitations of the PDSI, the index has been widelyadopted and used for a number of climatological stud-ies (Alley, 1984).

Seasonality of WildfireWildfire in the western coterminous U.S. is stronglyseasonal, with 95% of fires occurring between Mayand October. Fire starts peak during July and August.Depending on location, 50 to 80% of the western U.S.precipitation occurs between October and March. Bycontrast, the peak of fire season occurs during the hot-test and driest portion of the climatological annualcycle.

In conjunction with the hottest and driest time of year,monthly mean acres burned also peaks in July and Au-

gust, but shows somewhat different spatial featuresthan do fire starts. The areas with the largest numberof acres burned tend to be in regions of smaller fueltypes (e.g., grasses, shrubs, chaparral), though not exclu-sively. Smaller fuels typically lose moisture more rapidlythan heavier fuels, increasing their fire consumption po-tential. These same regions tend to be climatologicallywindy areas, such that once a fire starts, the combina-tion of fuel factors and wind cause rapid spread.

The progression of the fire season varies geographically.The fire season develops earliest (May and June) andends earliest (August) in New Mexico and Arizona.The start of the fire season spreads north and westthrough July and August. To the north, the fire seasonin northern Idaho and western Montana is more con-centrated toward the later part of the summer, withroughly 50% of fire starts occurring in the hottestmonth, August. The fire season in California peaks inSeptember, aggravated by hot, dry conditions thatbuild through the summer.

Examples of Links Between Climate and WildfireSeason Severity: Sierra Nevada and Great BasinGiven that weather, fuels, topography and human in-tervention all affect fire in varying degrees, one needsto be cautious in attributing the number of acresburned to individual factors. It is an over-simplifica-tion to claim that a large fire was caused solely by lowhumidity or no precipitation. The question then be-comes, to what extent do certain climate factors affectwildfire characteristics? A primary role of climateseems to be affecting vegetation conditions favorablefor ignition and spread, in addition to determining thefrequency and location of ignition by lightning. Ourresearch demonstrates that moisture anomalies can ex-ert a strong influence on fire severity over large spatialscales and long lead times.

Lagged correlations for August fire starts and acresburned with lagged, prior-season divisional PDSIscores (Figure 1) show interesting regional relation-ships. In the Sierra Nevada (Figure 1a), anomalous firestarts are negatively correlated with PDSI scores fromthe preceding year and a half; that is, increased num-bers of fire starts in August are associated with deficitPDSI in the prior winter and spring, and concurrentsummer. August anomalous acres burned in the SierraNevada appear to be associated with a deficit in PDSIin spring and summer. This suggests that during thespring and summer of the previous year, moist condi-tions are conducive to the growth of some fine fuels,

27


while prolonged dryness in the nearer term increasesfuel flammability through vegetation mortality, loss ofvegetation moisture and duff dryness. Thus, in the Si-erra Nevada, fuel loads are relatively more importantfor acres burned than for fire starts, and fuel flamma-bility is more important for the number of fire starts.

In the Great Basin, by contrast, the correlation be-tween August fire starts and acres burned is muchhigher. Figure 1b indicates that deficit PDSI in springand summer is not very important to either anomalousfire starts or acres burned in August. Much of the re-gion is comprised of grasses, which follow an annualcuring cycle providing a readily available fuel sourcefor fire, thus an anomalous precipitation deficit mighthave little impact. Invasion of grasses, in particular byexotic species, have increased fire occurrence across theGreat Basin during the past few decades. The seasonalgrowth and development of these fuels are strongly af-fected by precipitation anomalies. Large positive corre-lations with PDSI 12 to 15 months prior to August(Figure 1a) suggest that anomalous precipitation af-fects the previous season’s fine fuel production, andthus increases the current season’s fuel load. The corre-lation between August acres burned and previous yearPDSI is higher in the Great Basin (Figure 1a) than inthe Sierra Nevada (Figure 1b). We believe that fire dy-namics, especially, large acreage fires, in the Great Ba-sin are dominated by fine fuels, whereas the Sierra Ne-vada has both fine fuels and heavy fuels. The heavy fu-els are slower growing and burn less frequently thanthose in the Great Basin. This suggests that one year’sprecipitation is not as important in determining thefuel load in the Sierras as it is in the Great Basin.

It is important to reiterate that fire starts and acresburned are not determined solely by climate and veg-etation, but despite idiosyncracies in human interven-tion and recording, these our studies tell us a great dealabout how climate forcing affects fire risk in differentlocations.

Western Wildfire Seasonal ForecastThe links between seasonal climate anomalies and sea-sonal fire activity in the Western US prompted us toforecast seasonal acres burned (May to October) on a1° x 1° grid using lagged values of PDSI (Westerling etal., 2001b). The forecast model is estimated usingprincipal components (PC) regression to calculate lin-ear relationships between PCs of the seasonal acresburned and lagged PDSI. Acres burned per grid cellwere summed for fires starting between May 1 and

Figure 1. Correlations of PDSI with August Fire Activity inthe (a) Sierra Nevada, (b) Great Basin.

Figure 2. Map of forecast model skill. Yellow areasindicate where the model has little skill and red to darkbrown areas indicate where the model has good skill. Thisfigure shows the correlation of jacknifed, cross-validatedforecast anomalous acres burned with actual anomalousacres burned, 1980-2000.

28

October 31 and scaled using a log10

transformation.The 312 grid cells averaging more than one fire peryear comprise the predictand data set. For predictors,110 western U.S. Climate Division PDSI series areused at 7 different lags: January and March immedi-ately preceding, January, March, May, and August oneyear previous to, and May two years prior to the fireseason, for a total of 770 predictor variables. The di-mensions of the predictor and predictand data setswere reduced by substituting the first 8 PCs for each ofthe two data sets.

Forecast skill is measured here by the correlation be-tween cross-validated model output and thepredictand—log

10 of seasonal acres burned—for each

grid cell (Figure 2). While this model was not opti-mized for any particular region, the map in Figure 2shows the greatest skill in the Rocky Mountains, theSierra Nevada, central Arizona, and the Great Basin.Note that clear areas can indicate either no data or noskill.

A forecast, developed retrospectively, for the very activefire season of summer 2000, was fairly successful in re-producing the observed acres burned. Cross-validatedforecast anomalous acres burned for the 2000 fire sea-son (Figure 3a) show a similar spatial pattern in signand intensity to the actual anomalies (Figure 3b). Con-sidering that the 2000 fire season was an extreme yearin many locations compared to the previous 20-yearrecord used to estimate the model, this result stronglyindicates the utility of this approach to forecasting thewestern US wildfire season.

Finally, the 2001 fire season was predicted using asimilar set of lagged PDSI predictors. The 2001 fireseason forecast (Figure 4) uses persistence in the Febru-ary 2001 PDSI to model March 2001 PDSI; otherwisevariable definitions are the same as for the 2000 fore-cast. Note that the forecast, while exhibiting positiveanomalies in an arc from eastern Washington throughthe Rockies and New Mexico, seems to indicate amuch less extreme fire season than in 2000.

ConclusionsGridded numbers of wildfire starts and acres burnedfrom the BLM, USFS and BIA can be used to charac-terize the seasonal and interannual evolution of fireseasons over the last two decades. These data show im-portant relationships between fire season severity andcurrent and previous years’ climate. Acres burned indry shrub and grasslands as in the Great Basin appear

Figure 3. Forecast number of acres burned for May-October 2000 (a); actual number of acres burned for theMay-October 2000 forecast (b).

Figure 4. Map of forecasted acres burned for May-October 2001. This forecast was made early in the springof 2001. Forecasted acres burned are shown asanomalies, thus orange to red areas indicate an aboveaverage number of forecasted acres burned and green toblue areas show a below average number of forecastedacres burned.

29


to depend strongly on fuel accumulation governed byclimate conditions 10-18 months before the fire season(in addition to annual carryover fuel), and may be rela-tively unaffected by contemporaneous climate. In theSierra Nevada range, central Cascades, and NorthernRockies, fire season severity is negatively correlatedwith contemporaneous PDSI and positively correlatedwith PDSI from the previous summer. This result re-flects a trade-off between fuel accumulation and flam-mability, with wet conditions the previous year con-tributing to fuel accumulation and wet conditions inthe current year suppressing fire activity.

The relationship between ENSO phases and fire startslacked statistical strength. Correlations with climateindices at scales of one season to a year or longer leadtimes indicate that fire season prediction is possible ona similar scale. It is commonly assumed that fire seasonseverity depends on precipitation and temperatureconditions earlier in the year; however, our resultsdemonstrate regional variability in the importance ofantecedent seasons’ precipitation for determining fireseason severity. In many locations the climate condi-tions with the greatest relevance for future fire seasonseverity occur at lead times of one year or greater.

Our 2000 wildfire season severity hindcast showed suf-ficient skill, we believe, to indicate that future forecastscan be used to guide fuel management and resource al-location decisions. Our 2001 forecast indicates greaterthan average fire severity in an arc from eastern Wash-ington through the Rockies and New Mexico, a muchless severe fire season in the Mojave and Great Basin,and only a marginally more intense season in the SierraNevadas compared to last year’s prediction. A wide va-riety of choices for predictor variables and model speci-fications remain to be explored.

ReferencesAlley, W. M., 1984: The Palmer Drought Severity In-dex: limitations and assumptions. J. Climate Appl. Me-teor., 23, 1100-1109.

Westerling, A. L., Brown, T. J., Gershunov, A., Cayan,D. R., Dettinger, M.D., 2001a, Climate and Wildfirein the Western United States, submitted to Bulletin ofthe American Meteorological Society January 2001.

Westerling, A. L., Cayan, D. R., Gershunov, A.,Dettinger, M.D., Brown, T. J., 2001b, Statistical Fore-cast of the 2001 Western Wildfire Season Using Princi-pal Components Regression. Experimental Long-Lead

Forecast Bulletin (http://www.iges.org/ellfb)

Related ResourcesCalifornia Applications Program (CAP) Web Pagehttp://meteora.ucsd.edu/cap

Western Wildfire Season Forecasthttp://meteora.ucsd.edu/%7Emeyer/fire_forecast.html

Climate Prediction: ENSO vs Non-

ENSO Conditions

Douglas Le Comte (NWS/NCEP/Climate PredictionCenter)February 15, 2001 9:00 AM

NOAA Climate Prediction Center (CPC) Long-leadForecast Tools• Coupled Model (CMP)—Ensemble mean forecast

of a suite of 20 GCM (general circulation model)runs forced with tropical Pacific sea surface tem-peratures (SSTs), produced by a coupled ocean-at-mosphere model. Available for leads of 1-4months.

• Canonical Correlation Analysis (CCA)—Predictspatterns of temperature and precipitation based onpredictor patterns of global SSTs, atmospheric 700mb heights, and U.S. surface temperature and pre-cipitation from the past year.

• ENSO composite—Supplies historical frequenciesof three forecast classes (above, within, below the1961-1990 mean — this will change to 1971-2000 mean in May 2001) for past years whenEquatorial Pacific SSTs indicate moderate orstrong El Niño or La Niña conditions.

• Optimal Climate Normal (OCN)—Predicts tem-perature and precipitation based on persistence(trends) of observed average anomalies for tem-perature (past 10 years) and precipitation (15years).

• Constructed Analog on Soil Moisture (CAS)—Constructs a soil moisture analog from a weightedmean of past years. Proportional weights are usedfor temperature and precipitation.

30

• Screening Multiple Linear Regression (SMLR)—Uses the same fields as CCA, but these predictorsare used to predict conditions at single stations.This method also uses soil moisture as a predictor.

Forecast SkillIn terms of seasonal skill, most forecasting tools do notwork well in late spring (April-May) and summer. Thisis especially true for NOAA CPC precipitation out-looks. Due in part to its stronger spatial coherence,temperature is easier to predict than precipitation.NOAA CPC 3-month temperature outlooks havesome skill, with considerable skill mainly during win-ter (Figure 1). Although precipitation skill is, in gen-eral, lower, precipitation skill is relatively high in win-ter and high during El Niño events (Figure 2).

Specific forecast tools include:

• CAS. The analog forecast from soil moisture,which combines data from 1932-1997, can beused to forecast temperature and precipitation forthe next few seasons. Seasonally, the highest skillfor temperature forecasts is from April-September,with peak skill for early summer forecasts. For2001, the CAS (based on February 7, 2001 condi-tions) forecast for spring (March-May) shows be-low-average precipitation in the Southeast andwetter than average conditions over the southernPlains and Pacific Northwest. The temperatureoutlook for spring shows above-average warmthover the Rocky Mountain states (Figure 3). CASsoil moisture forecasts show negative soil moisture

Figure 1. Heidke skill score summary for NOAA CPC 3-month temperature outlooks. Skill is aggregated over theentire U.S. And is estimated for forecasts issued 0.5months in advance of the season of interest.

Figure 2. Heidke skill score summary for NOAA CPC 3-month precipitation outlooks. Skill is aggregated over theentire U.S. And is estimated for forecasts issued 0.5months in advance of the season of interest.

Figure 3. Constructed analog (CAS) temperature (top)and precipitation forecasts for spring 2001 (March-May)from soil moisture data.

31


anomalies in the Southeast and positive soil mois-ture anomalies across the southern Plains statescontinuing through the summer.

• ENSO Composites based on CPC Soil Model. Thistool is also used to monitor drought; 5,000-6,000temperature and precipitation station reports areinput each day and, unlike the Palmer drought se-verity index (PDSI), it is updated daily. The CPCSoil Moisture Composite demonstrates some in-teresting patterns with regard to ENSO (Figure 4).In July preceding La Niña winters, there is a ten-dency for negative soil moisture anomalies acrossthe northern tier of the country and positive soilmoisture anomalies across the Southwest, espe-cially New Mexico. During neutral Julys, the ten-dency is for the Southwest and central Plains to bedry. During El Niño Julys there is a tendency forthe Southwest to be dry (except New Mexico),whereas the northern Plains, Texas and Floridahave positive soil moisture anomalies.

• OCN. The most skillful seasons for OCN precipi-tation forecasts are September-November throughJanuary-March. In contrast, the other tools tend tobe most skillful for late winter forecasts.

• CCA. The best skill for this tool is for winter tem-perature and precipitation forecasts and summertemperature forecasts.

• SMLR. Unlike CCA, this tool shows great skill forfall forecasts.

• CMP. Coupled model skill appears to be heavilydependent on ENSO extremes.

• NOAA Seasonal Drought Outlooks. These takelarge-scale trends based on subjectively derivedprobabilities guided by numerous indicators, in-cluding soil moisture forecasts, extended rangeforecasts, long-lead outlooks, and other tools toforecast drought trends over the next 3_ months.The outlook through April 2001 shows improve-ment likely over the northern Rockies, the centralPlains, West Texas and most of the Southeast, butdrought is likely to persist in Florida.

ENSO vs. Non-ENSO Forecast SkillSkill for official CPC climate outlooks varies with sea-son, variable and phase of ENSO. Temperature out-looks for the lower 48 states are most accurate for late

Figure 4. NOAA CPC July soil moisture composites. A. LaNiña conditions. B. Neutral conditions. C. El Niñoconditions.

32

Figure 5. Comparison of (top) NOAA CPC PrecipitationOutlook for January-March 1998 with (middle) observedprecipitation rankings in the wettest or driest 5 years(since 1895), and (bottom) a composite of 13 20th centuryEl Niño events. Green indicates wet conditions and red/brown indicated dry conditions.

winter and late summer; they are least accurate for latespring and late fall. Precipitation forecasts are generallyless skillful than temperature forecasts, with marginalskill for all tools even in their best seasons and loca-tions under “normal” circumstances. El Niño givesforecasters a real advantage. For example the January-March 1998 forecast was exceptionally accurate (Fig-ure 5). During strong El Niños or La Niñas, precipi-tation skill can be as high as temperature skill for coolseason forecasts. Areas of enhanced skill include thesouthern third of the country, the northern Rockies,the High Plains, and the Ohio Valley. Strong La Niñaconditions imply the possibility of moderate precipi-tation skill for some parts of the warm season as well,especially for the Southeast. Note that part of theproblem with precipitation forecast skill is due toproblems with snow and rainfall data. Mountain re-ports and high elevation stations are discarded in fa-vor of longer-term records from the cooperative net-work. There are discussions between United StatesDepartment of Agriculture (USDA) and the National

Weather Service (NWS) on upgrading the cooperativenetwork, and SNOTEL reports from western moun-tain areas are now included in a unified CPC rainfallanalyses.

An examination of Heidke skill scores shows the fol-lowing:

• Skill is correlated with ENSO. Precipitation fore-casts do improve with ENSO conditions and thereis a significant correlation between precipitationskill and an ENSO index (the MEI; Wolter andTimlin, 1993),

• During El Niño, precipitation skill scores have beenpositive for 12 forecasts and negative for 3, whereasduring La Niña, precipitation skill scores have beenpositive for 12 forecasts but negative for 9.

• Temperature skill scores are high in winter, regard-less of ENSO conditions.

In summary, CPC outlooks for temperature and pre-cipitation are best in winter. The best precipitationforecasts are during El Niños, especially for non-CL ar-eas (i.e., regions for which forecasts are ventured).Nevertheless, there is hope in summer that forecastsbased on antecedent soil moisture will lead to improve-ments.

U.S. Drought MonitorA brief word about the weekly Drought Monitor. TheDrought Monitor is an expert consensus based onmany different forecast tools. The principal inputs forthe Drought Monitor are the CPC daily soil model,PDSI estimates, United States Geological Survey(USGS) streamflow reports, the most recent 30-dayprecipitation, USDA soil ratings and the satellite veg-etation health index. The satellite vegetation health in-dex monitors growing conditions using informationfrom visible, near-infrared, and thermal bands. It isbased on anomaly data back to 1985. The DroughtMonitor is produced by three interagency partners,NWS/CPC, USDA/NOAA Joint Agricultural WeatherFacility (JAWF), and the National Drought MitigationCenter (NDMC). A fourth agency, the National Cli-matic Data Center, will be added in the spring of2001. Numerous outside experts, including the USGS,state climatologists, regional climate centers and NWShydrologists, provide valuable feedback that is used inthe final weekly product. It is posted on the Internetevery Thursday morning.

33


Climate Prediction Interpretation and

Evaluation Workshop

Holly Hartmann, Thomas Pagano and SorooshSorooshian (Department of Hydrology and WaterResources, University of Arizona)February 15, 2001 10:00 AM

IntroductionThe goals of this workshop were to examine monthlyand seasonal climate forecasts. In order to meet thesegoals the workshop will include the following: (1) asurvey of the different types of forecasts available, (2)explanation of the correct interpretation of climateforecasts and a hands-on exercise in forecast interpreta-tion, (3) a discussion of forecast evaluation methods(including a hands-on forecast evaluation exercise anddiscussion), and (4) some examples of evaluations ofofficial government climate outlooks, with an empha-sis on evaluations that are useful for fire management.The latter two aspects, part of research in progress onforecast evaluation products to be released later thisyear, have been omitted from the following brief sum-mary of the workshop.

Climate ForecastsVarious branches of the National Oceanic and Atmo-spheric Administration (NOAA), most prominentlythe National Weather Service (NWS) have been issu-ing monthly and seasonal weather and climate out-looks for over 30 years. Beginning in the mid-1990sthe NOAA Climate Prediction Center (CPC) has beenissuing seasonal climate outlooks, consisting of proba-bilistic temperature and precipitation forecasts, morethan one season in advance. In addition, other agen-cies, including the Research Institute for Climate Pre-diction (IRI) and private forecast providers have beenissuing seasonal climate forecasts for the past severalyears. The official source for seasonal climate forecasts

is NOAA/NWS, and their work will be the focus ofmuch of this workshop.

Forecast types include categorical, deterministic, andprobabilistic seasonal outlooks. Categorical outlooks fore-cast whether future conditions will fall into a particularcategory, such as normal, above normal, below normal.These forecasts can literally be interpreted as saying thatthere is a 100% chance that forecasted category will oc-cur and a 0% chance that any other category will occur.Forecast quality is judged by whether or not the ob-served parameter fell into the forecasted category. Cat-egorical outlooks were issued by NOAA prior to 1983.

Deterministic outlooks are framed in terms of a specificquantity, e.g., 4 inches of rainfall or 60% of normal.Such forecasts are intuitively appealing and easy to usein planning. Forecast quality is judged by how closethe observed is to the forecasted quantity. The forecastvalue, however, is unlikely to ever exactly equal theoutcome. Deterministic forecasts often lead to unreal-istic ideas about forecast confidence, especially whenthey neglect confidence bounds.

Probabilistic forecasts have been issued by NOAA sincethe early 1980s. Probabilistic forecasts indicate theprobability of the forecast observation being in a cer-tain category, e.g., a 55% chance of precipitation fall-ing in the wettest 1/3 of the historical record. NOAApresently forecasts the probability of parameters fallinginto three categories (i.e., terciles) determined by rank-ordering observed values for a recent 30-year period(10 values in each category). Evaluation of probabilis-tic forecasts is also troublesome because they alwaysgive every category some chance of happening. TheIRI also issues probabilistic seasonal climate forecasts,with predicted outcomes expressed in terms of terciles.

Future forecast tools, so-called next generation seasonaloutlooks, are expressed in terms of probability ofexceedance. These forecasts express information aboutthe entire range of possibilities (not just terciles as inthe NOAA/CPC seasonal probabilistic outlooks).Moreover, probability of exceedance forecasts provideprobabilities and quantities for individual locations.These forecasts are often difficult to understand andapply, however they contain much more informationthan any of the previously available forecast formats.

Interpretation of Climate ForecastsIn order to understand how to interpret forecasts, it isimportant to know (1) what the forecast provider con-

Related ResourcesNOAA CPC Seasonal Climate Outlooks:http://www.cpc.noaa.gov/products/predictions/90day/

U.S. Drought Monitor:http://enso.unl.edu/monitor/monitor.html

U.S. Seasonal Drought Outlooks:http://www.cpc.noaa.gov/products/expert_assessment/seasonal_drought.html

34

siders normal, (2) how probabilities are expressed inmaps and legends, and (3) how to identify and inter-pret non-forecasts. CPC and IRI currently base normalon 1961-1990 data; later this year, normal will bebased on 1971-2000 data. Normal is not a single value,but the range of values experienced in the 10 most-av-erage years in the 30-year period. Note: this methoddoes not represent the range of variability in the entireperiod of historical record (which, in the Southwest, is~100 years long), which may be much greater andshould not be forgotten.

The CPC method for expressing changes of probabil-ity in their climate outlooks is as follows: CPC dividesthe normal period (1961-1990) data into terciles andindicates probabilities on their climate outlook mapsin terms of the increase in the probability of being in thetop or bottom tercile. Thus, a value of 5% in an area of“A” (excess likelihood of above normal) on a precipita-tion outlook map corresponds to:

• 38% chance of precipitation in the highesttercile of the 1961-1990 data distribution(i.e., 33% + 5%)

• 33% chance of precipitation in the middletercile of the 1961-1990 data distribution(i.e., probability unchanged)

• 28 % chance of precipitation in the lowesttercile of the 1961-1990 data distribution(i.e., 33% - 5%)

The designation CL on NOAA/CPC seasonal probabi-listic outlooks does not mean climatology or normal.Rather it indicates insufficient skill to issue a forecastor disagreement among individual forecast techniques.Where a CL rating covers much of U.S. a managerwould do well to consult more regional or localizedforecasts, which may have more skill. Where no fore-cast exists with adequate skill, the conservative ap-proach is to avoid actions that assume any particularclimate condition (i.e., respond more to current condi-tions while being prepared for any event). The impor-tant distinction is not to misinterpret “unknown/notpredicted” (CL) as “normal”.

The Results of Hands-On Forecast InterpretationExercisesThis survey investigated how fire managers perceiveand interpret the current generation of climate fore-casts. In particular, it focused on the forecast format,forecast probabilities, the interpretation of “climatol-ogy” (CL) designations on forecast maps, and the op-

erational definitions of terms like dry and wet.

MethodologyAs part of the workshop on forecast evaluation, clima-tologists and fire managers were asked to write onnotecards their responses to questions about forecastinterpretation. This helped to focus discussion on cer-tain aspects of the CPC forecast format (i.e., CL fore-casts). At the beginning of the workshop, participantswere asked to divide into working groups according tothe following categories:

• Climatologists/Meteorologists• Fire planning• Fire operations• Other (e.g., fire research).

This was done to prevent climatologists from influenc-ing the responses of fire managers, and to allow firemanagers an opportunity to interact with peers.

In order to ascertain how well climatologists “knowtheir mate”, they were asked to respond in terms ofhow they thought the fire managers would respond,instead of in terms of what they personally believed tobe the correct answer. Every time a new question wasasked, the climatologists were reminded to answer inthis way. The results of this survey were briefly summa-rized at the beginning of the February 15, 2001 after-noon session and are analyzed more in depth here. Thereader should keep in mind that these answers are notrepresentative of the broader fire management commu-nity – workshop attendees were clearly interested inclimate and have experience in interpreting forecasts.

CPC forecasts had been presented and discussed fre-quently during the sessions prior to this workshop. In-deed, the two hours of presentations immediately priorto the workshop focused specifically on the most re-cent CPC forecasts in the context of the coming fireseason (see presentations by LeComte, Wolter and oth-ers elsewhere in this volume). Members of the “other”category (fire research) had the table farthest in theback and may not have been able to read the smallerprint on the slides about IRI forecasts. There weremany answers left blank on the notecards of these re-spondents.

Interpretation of Forecast ProbabilitiesQuestion 1: While looking at a CPC forecast map for pre-cipitation for the winter 2000, participants were asked towrite down their answer to the question “What is the

35


Figure 1. Seasonal climate forecast issued by the National Weather Service Climate Prediction Center and used in thehands-on forecast interpretation exercises.

forecast for Tucson, Arizona?” The forecast legend wasnot shown, because it is often left out when CPC fore-casts are communicated by other groups. The map, asshown in Figure 1, indicated that there was a 53-63%probability that Tucson would be in the driest tercileof the 1961-1990 historical record, a 33% probabilityit would be in the middle tercile and a 3-13% chance ofbeing in the wettest tercile of the record. The responseswere categorized into 6 different classes as follows:

• Categorical and “correct” (e.g. “warm anddry”, “below normal precip”, “dry, less precipi-tation”)

• Categorical and “incorrect” (e.g. “above aver-age precip” – the forecast was for an enhancedprobability of dry)

• Probabilistic and “correct” (“greater likelihoodof dry”, “48% chance it will be drier than nor-mal”1)

• Probabilistic but “incorrect”• Deterministic (“48% of normal”)• Other (“wetter than normal temps”, “drier by

3 units of some kind”)

In Table 1 the correct response is highlighted in gray.In this table, as in all of the following tables, the num-bers shown represent the number of responses for eachcategory, among each class of users. The total numberof users in each class is given in parentheses at the topof each table. For example, Table 1 under “Planning,”

6 planners participated and 5 responded with a cat-egorical “correct” answer and 1 responded “other”.

Interpretation of CL forecastQuestion 2: While looking at a CPC forecast map for pre-cipitation for the winter 1999-2000, participants wereasked to write down their answer to the question “What isthe forecast for Reno Nevada?” The forecast legend wasalso omitted for this question. The map (Figure 1)showed this area had a CL forecast, which can be alter-natively interpreted as saying complete lack of forecasterconfidence or all conditions equally likely. We have classi-fied the answers as follows:

• Normal (e.g. “near normal,” “normal”)• Climatology (e.g. “no skill,” “no forecast,”

“climatology”)• Other (“nothing significant forecasted,”

“slightly above normal”)

The Effect of Making the Forecast Legend AvailableQuestions 3 and 4: After making the CPC legend avail-able, participants were asked questions one and twoagain. Participants responded in the following ways:

1 This probability may not match with the actual probabilitybecause the CPC contour intervals are uneven; that is thefirst two contours have intervals of 5% and deepercontours have intervals of 10%. See legend for details.

36

”?ZAnoscuTroftsacerofehtsitahW“:dnegelehtgnieesretfA.3elbaT

:fosmretnidemarfesnopseRgninnalP

)6=n(snoitarepO

)7=n(rehtO)8=n(

latoTeriF

)12=n(

&stsigolotamilCstsigoloroeteM

)51=n(

*tcerrocydaerlarewsnA 0 1 1 2 2

rewsnatcerrocnitpeK 3 1 5 9 7

otlacirogetacmorfdegnahC*citsilibaborp

1 3 1 5 4

rehtO 2 2 1 5 4

.sesnopserderreferP*

”?oneRroftsacerofehtsitahW“dnegelehtgnieesretfA.4elbaT


)6=n(snoitarepO

)7=n(rehtO)8=n(

latoTeriF

)12=n(


)51=n(

*tcerrocydaerlarewsnA 0 1 3 4 7

rewsnatcerrocnitpeK 4 2 3 9 8

*ygolotamilcotdegnahC 1 2 0 3 1

rehtO 1 2 2 5 1

.sesnopserderreferP*

”?ZA,noscuTroftsacerofehtsitahW“.1elbaT


)6=n(snoitarepO

)7=n(rehtO)8=n(

latoT*eriF)12=n(


)51=n(

tcerrocdnalacirogetaC 5 6 6 71 01

tcerrocnidnalacirogetaC 0 0 0 0 1

**tcerrocdnacitsilibaborP 0 1 1 2 2

tcerrocnitubcitsilibaborP 0 0 0 0 0

citsinimreteD 0 0 0 0 0

rehtO 1 0 1 2 2

.stsigolotamilc-nonllastneserpereriflatoT*.rewsnatcerroC**

”?VNoneRroftsacerofehtsitahW".2elbaT


)6=n(snoitarepO

)7=n(rehtO)8=n(

latoTeriF

)12=n(


)51=n(

egarevA/lamroN 5 3 4 21 8

*ytniatrecnulatoT/ygolotamilC 0 1 3 4 7

rehtO 1 3 1 5 0

.rewsnatcerroC*

37


• They kept their previous correct answer• Kept their same, but incorrect, answer• Changed from a categorical to a probabilistic

forecast (e.g., “53-63% chance of below nor-mal,” “10-20% chance of below normal,”“there is a 10-20% probability anomaly”

• Gave some other answer, usually mistaking aprobability anomaly for a “percentage of nor-mal” forecast (e.g., “10-20% below normalprecip,” “below,” “43.3-53.3% [no units],”“precipitation 50% below normal”)

Definitions of “Dry” and “Wet”Question 5: Using the same CPC map, participants wereasked to compare the forecast for Tucson and the forecast

for central Florida and were asked to answer the question“Which do you think will be wetter, Tucson or centralFlorida?” The probability anomalies were stronger inFlorida than in Tucson. However, Florida is generally awetter place. The correct answer is that one does notknow. The CPC maps cannot answer the question fortwo reasons (1) They are not forecasts for a quantity ofrainfall (they forecast the probability of precipitationfalling in one of three categories), and (2) the defini-tion of dry is relative to the location and season. Therewere four types of answers:

• Arizona will be wetter• Florida will be wetter• Neither/Can’t tell

"?adirolFronoscuT,retteweblliwhcihw,tsacerofsihtotgnidroccA“.5elbaT

esnopseRgninnalP

)6=n(snoitarepO

)7=n(rehtO)8=n(

latoTeriF

)12=n(


)51=n(

adirolF 0 4 2 6 4

anozirA 4 2 2 8 7

*rehtieN 1 1 4 6 4

rehtO 1 0 0 1 1

.esnopsertcerroC*

?denifed”yrd“siwoh,skooltuolanosaesCPCehtottcepserhtiW.a6elbaT


)6=n(snoitarepO

)7=n(rehtO)8=n(

latoTeriF

)12=n(


)51=n(

lamroNnahtsseL/naeMwoleB 1 6 3 01 01

*selicretotecnerefeR 0 0 0 0 1

mrofrehto,noitatipicerpsseL 4 1 3 8 4

rehtO 1 0 2 3 0

.esnopsertcerroC*

?”yrd“enifedIRIseodwoH.b6elbaT


)6=n(snoitarepO

)7=n(rehtO)8=n(

latoTeriF

)12=n(


)51=n(rofsarewsnatcerrocni,emaS

CPC4 1 4 9 7

*selicretotecnerefeR 0 0 0 0 1

dnaCPCmorftnereffiDtcerrocni

1 6 1 8 5

rehtO 1 0 3 4 2

.esnopsertcerroC*

38

• Other (e.g., one respondent gave a probabilis-tic forecast for Florida)

Question 6a: What is the CPC definition of “dry”?. CPCtakes a 30-year historical period, ranks those yearsfrom wettest to driest and then divides that list into 3parts with 10 years in each category. The wettest 10years are considered “wet”, the driest 10 years are con-sidered “dry” and the other 10 are considered normal.When CPC says there is a higher probability of dryconditions, they define dry by using the driest 10 yearsof the 30-year record. Workshop participants’ re-sponses to the question fell into four categories, inwhich “dry” is defined as:

• Below mean/Less than normal, with no otherquantities identified

• Location in lowest tercile (“lowest third”)• Less precipitation, but in some other form

(e.g., “10-20%”,”lesser degree of precipita-tion”, “predicting no precipitation”, “drier hasmeaning, dry has none”, “5-10% below nor-mal”)

• Other (e.g., blank, “don’t know”)

Question 6b: Participants were asked to answer questionsabout an IRI forecast. One of the questions was “Howdoes IRI define “dry”?” IRI and CPC have the same defi-nition of dry, although their forecast map formats aresignificantly different. One climatologist inquired aboutwhether we were referring to the terciles or IRI’s specialdesignation of “dry season”, which appears as a grayedout “D” area on the IRI maps. “Dry season” means thatless than 15% of the annual precipitation usually fallsduring this particular 3-month period. We clarified toall participants that we were not referring to the “D” onIRI maps. There were three categories of answers:

• The same as the answer given for CPC, butincorrect (i.e., they answered “below mean”)

• the correct interpretation, which is “in thelowest tercile”

• A different answer than the one given forCPC, but still incorrect (“lack of moisture?”“less than 75% of normal,” “8% or moreprobability of being dry,” “dry outside meanwith confidence interval”)

• Other (blank, “don’t know”)

Summary and Conclusions of Workshop SurveyThe current probabilistic CPC seasonal climate fore-cast format represents a great improvement in informa-

tion content over the simplified “wetter than normal/drier than normal” forecasts of the 1970’s and earlier.These forecasts convey forecaster confidence and un-certainty in ways that categorical forecasts cannot.High probability anomalies indicate that forecasters arehighly confident in their predictions that the observedwill be in a particular tercile of the historical record.On the other hand, CL (the white areas on the CPCmonthly and seasonal outlook maps) indicates a com-plete lack of forecaster confidence – perhaps the mostimportant forecast of all.

The results of the survey show that fire managers donot view the current generation of forecasts in probabi-listic terms. The most common misinterpretation ofthe CPC forecasts is that the contours indicate the se-verity of drought/wetness expected. This is not true,especially in the case of CL, which was often misinter-preted as a forecast for normal conditions. Contoursshowing weak probability statements (5-10% probabil-ity anomalies) can be dangerous, especially when theyare the strongest statement on a map. We found thatusers may not interpret the forecast for their region,but rather in terms of their region relative to the rest ofthe country for a given forecast. This problem is par-ticularly relevant during non-ENSO conditions.

Given that almost all fire managers interpreted theCPC outlooks as categorical forecasts, categorical fore-cast evaluation measures (i.e., Probability of Detection,False Alarm Rate) may be an appropriate and under-standable entry point into forecast evaluation. How-ever, probabilistic evaluation scores should also be usedto remind forecast users of the probabilistic informa-tion contained within (otherwise, if they only see cat-egorical evaluation measures, they might think theforecasts are categorical).

Finally, there were divergent interpretations of CPC’sdefinition of fundamental terms such as wet, dry andnormal. The base period (1961-1990) rarely enteredinto the discussion when a CPC forecast was pre-sented. Likewise, some users may not have been awarethat terciles were being used; many interpreted theforecasts as above-normal or below-normal withoutreference to the middle category (given the fixed prob-ability in the “normal”/middle category, this misinter-pretation is relatively benign). However, users thathave more extreme definitions of dry (e.g., “predictingno precipitation”) have a greater potential to be disap-pointed by the forecasts. Although they were not dis-cussed, recent temperature forecasts have been strongly

39


influenced by frequent warm conditions throughoutthe 1990’s. For example, a forecast for warm condi-tions may have a different meaning to someone whohas only experienced the climate of the past 5 years asopposed to all of 1961-1990.

In summary, fundamental work on forecast format andcommunication remains to be done. This effort is notonly the responsibility of the forecasters, but of the us-ers as well. Before a user bases a decision on a forecast,it is extremely important for them to understand theprecise (albeit non-intuitive) interpretation of the fore-cast. Likewise, it is often just as important to knowabout the past as it is to know about the future. Firemanagers could benefit from increased exposure to his-torical climate data, in particular the baseline period ofthe climate forecasts.

Clearly the utility of the climate forecasts is enhancedwhen forecast users and producers share a commonlanguage about the forecasts, their uncertainty andtheir application. However, it is probably not realisticto expect all fire managers to become versed in clima-tology or forecast producers to become versed in firemanagement. We believe that both communitieswould benefit from an intermediary group — person-nel versed in both the language of the forecasters andthe users, in order to facilitate the transfer of informa-tion between groups. CLIMAS, the program for Cli-mate, Ecosystem and Fire Applications and/or the vari-ous interagency coordination elements of the fire man-agement community may help to fill this role.

Related ResourcesNOAA CPC Climate Outlooks:http://www.cpc.noaa.gov/products/forecasts/

International Research Institute for Climate Prediction(IRI) Forecasts:http://iri.columbia.edu/climate/forecast/

40

Integrated Assessments

Interagency Integrated Assessments

Roger Pulwarty (NOAA Office of GlobalPrograms)February 16, 2001 8:00 AM

IntroductionWhat should an interagency approach to integrated as-sessment look like? The Regional Integrated ScienceAssessment (RISA) program is one approach. TheRISA approach is based on the concept of integrationamong disciplines, and between science and society.Such assessments seek to improve our understandingof climatic systems, climate variability, and the impactsof climate on human and natural systems through syn-thesizing and evaluating knowledge and projectionsgenerated from different sources and using an array ofdifferent disciplinary approaches. The end goal is todevelop decision support tools and climate services ap-plications.

The RISA program stresses the importance of rel-evance and practical problem solving. The objective isto characterize the state of knowledge about climateand its impacts, and to identify knowledge gaps thatcan be filled through answering carefully formulatedresearch questions. Through integrated assessment,syntheses of climate projections and evaluations can beproduced that incorporate physical and social scienceconsiderations, economic factors, and environmentalknowledge in a manner that improves society’s abilityto address and respond to climatic events, includingvariability and long-term changes. The process, in-volves developing new forms of integrated knowledgeby identifying social needs and risks. University scien-tists, the agencies that attempt to address social needsand risks, and the people who are affected act as col-laborators in the process. There is a two-way flow ofknowledge between these collaborators. In many casesit has been in the administrative arena where tradi-tional research efforts have failed: the questions may beknown, but remaining largely unknown is how tomanage large, complex systems. We cannot look atonly one sector in isolation.

Building and maintaining stakeholder relationships atlocal and regional levels is essential to developing such

knowledge and capacity. People need to know that re-search is ongoing, and is leading to better informationfor management and decision making. Further, the fo-cus needs to be on more than just preparedness andearly warning capabilities. Mitigation must be in-cluded as well.

The problem may be framed in terms of risk manage-ment (proactive, oriented toward mitigation and pre-paredness, framed around desire to improve predictiveand early warning capacity before disaster strikes) ver-sus traditional crisis management (reactive response af-ter an event such as a flood or drought has arisen).Public perception of hazards has usually arisen only af-ter development has occurred, resulting in a greater fo-cus being placed on reactive measures. Moving towardmore proactive approaches can be difficult, however;for example, strengthening land-use and buildingcodes may provide a proactive means of adapting toclimate variability, but may generate strong opposition.It is essential to identify who is doing what, and where,and to determine what they need. This can be com-plex. For example, 12 federal agencies deal with floodhazards; in addition, all 50 states have flood-prone ar-eas; there are 3000 flood control districts, and 20,000local governments having flood-prone areas!

Climate, RISAs and Fire ManagementAn Interagency Review and Update of the 1995 Fed-eral Wildland Fire Management Policy was issued thisyear (2001). The review reaffirms that what we knewin 1995 was basically correct, but now realization hasstruck that it is time to act. Conditions continue to de-teriorate, and are even worse than was expected in the1995 report. Fire in the urban/wildland interface hasbecome more complex and extensive. Likewise, ecosys-tem sustainability, issues regarding the use and role offire, and science applications, communications, andeducation have all become more immediate. Imple-mentation of the 1995 policy was incomplete due tothe quality of the planning, the degree of interdiscipli-nary coordination involved, limited interagency col-laboration, and criteria used for program evaluation.Also important was the need for program evaluationcriteria that cut across agencies, programs and disci-plines. Further, issues associated with public health,private property and infrastructure were not addressed.

41


We do not yet have a process for effectively incorporat-ing climate into fire management, although we dohave a good system for integrating weather into deci-sion-making. The problem, however, is not what to do,but how to do it. Cooperative research is needed toovercome this barrier. Joint development of tools toidentify, assess, and mitigate risks is required. The ulti-mate goal is to improve predictability and understand-ing of wildland fire-climate relationships before, dur-ing and after events. The societal values to be protectedneed to be defined, as well as long-range interagencyobjectives and optimal interagency preparedness levels.

In terms of policy strategy, the how of accomplishingthese tasks, we need to seek authorization to eliminatebarriers to funds transfers. We also need to developpartnerships with stakeholders and others, improvedata collection mechanisms, and standardize both thelanguage and process we use. It is important also tocommunicate effectively with property owners andothers. Among the challenges are determining how tocommunicate with the public about paradoxes, such asthat the same condition may be both a benefit and ahazard. How and when to move across scales from thevery local to national and international/global is anequally important challenge. Addressing contradictionsassociated with land use planning and management islikewise essential, particularly in cases where imple-mentation and enforcement of strong codes is neededbut strong opposition to such regulation exists.

Moving fire-climate initiatives forward requires localfeedback and support. How receptive are the agencies?Will such a push from the bottom up generate resis-tance at the top? Right now there is receptiveness to al-ternative approaches to fire management, so, I suggest,conditions may be more favorable than they have beenin the past for making progress.

Related ResourcesNOAA Office of Global Programs Regional IntegratedAssessments Programhttp://www.ogp.noaa.gov/mpe/csi/rgas/index.htm

Climate and Human Impacts on Fire

Regimes in Forests and Grasslands of

the U.S. Southwest

Barbara Morehouse (Institute for the Study of PlanetEarth, The University of Arizona) and Steve Yool(Department of Geography and Regional Develop-ment, The University of Arizona)February 16, 2001 8:30 AMMarch 28, 2001 11:20 AM

An interdisciplinary group of researchers at the Univer-sity of Arizona has recently been awarded a three-year,$1.26 million grant from the Environmental Protec-tion Agency (EPA) through its STAR Grant programto build an integrated GIS-based decision support toolfor use in fire management. The project team includes

• Barbara Morehouse, principal investigator andspecialist in social science surveys and institu-tional/policy analysis; Institute for the Studyof Planet Earth

• Thomas Swetnam, dendrochronologist andspecialist in fire history analysis; Director,Laboratory of Tree-Ring Research

• Steven Yool, biogeographer and specialist inGIS-based wildfire modeling; Department ofGeography and Regional Development

• Gary Christopherson, archaeologist and GISspecialist; Director, Center for Applied SpatialAnalysis

• Barron Orr, specialist in remote sensing; AridLands Center

• Jonathan Overpeck, climate specialist and Di-rector of the Institute for the Study of PlanetEarth

The integrated model will be built to provide decisionsupport for four specific areas: the Catalina-RinconMountains, Huachuca Mountains, Chiricahua Moun-tains, (all in Arizona) and the Jemez Mountains inNew Mexico (Figure 1).

The model is being structured to integrate physical/natural processes and human components into a singlesystem.

Research questions driving the project include:

• How might climate changes, in combinationwith changing land use patterns affect foresthealth, biodiversity, and ecosystem function?

42

• How might land use choices increase or de-crease ecosystem vulnerability to extremeweather events?

• What roles do human factors and behaviorsplay in elevating or reducing fire hazard?

• What roles do policies, organizational struc-tures, and communication patterns play inability to manage fire hazard effectively?

• What roles do public values and expectationsplay in fire hazard and in decision-makinglatitude and effectiveness?

The physical/natural process component is being con-structed using FARSITE as the foundation. FARSITEis a commonly used GIS-based fire hazard model thatallows fire managers to identify likely trajectories andintensities of fires based on an array of parameters suchas fuel moisture, topographic factors, vegetation char-acteristics, and so on. The existing data fields inFARSITE will be incorporated and expanded, whennecessary, into the new model. Among the data to beincluded in the new model are vegetation greenness,topography, vegetation and fire history, land covercharacteristics, soil characteristics, soil and vegetationmoisture. Climate data, including time series of pre-cipitation and temperature data, time series data onseasonal to interannual and decadal climate trends (in-cluding El Niño-Southern Oscillation [ENSO] datapatterns), and relative humidity data will also be incor-porated into this component of the model. Concurrentwith development of GIS-based data on natural andphysical processes, work is being carried out to build ahuman dimensions component. This component,which constitutes a major innovation in fire risk mod-

eling, includes data such as population numbers, den-sity, and distributions, land use patterns, roads andother access pathways, infrastructure, built environ-ment, economic data, and real estate values. Also in-cluded will be pertinent data derived from laws andpolicies, social values, and cultural history.

The construction of the physical-process componentmay be understood as a variation on the traditional firetriangle, wherein fuel, heat and oxygen combine tocreate a certain level of fire risk, into one where fire,climate and human factors combine to generate ecosys-tem risk. The integrated model will feature remote-sensing algorithms that link climate and vegetationmoisture in a manner that will produce a scaled assess-ment of impact and risk. Fuel moisture will be derivedby combining variables related to topography andgreenness. This entails use of high-resolution data thatreflect dynamics over time. Thus, fuel moisture statuswill be constructed at the pixel level. The model willprovide one-kilometer resolution, backed up by finer-scale fuel load data, at 60-meter resolution. Thus, thestandard fire risk model shown below

becomes inverted into this configuration:

The model will include current biogeography data,and data for the past twenty years, for each of thestudy areas; these data will cover soils, vegetation, fuelload factors, elevation, aspect, and so on. Correlationswill be made between climatic conditions over thistime period and specific fire-related events. The intentis to determine how these factors interact to produceparticular types and levels of fire and fire hazard.

firerisk

oxygen

heatfuel

ecosystemrisk

humanfactorsclimate

fire

Figure 1. Maps of study region and case study areas insoutheastern Arizona and north-central New Mexico.

43


The goals of the project are to improve our under-standing of the interactions among climate, fuel load,fire history and human factors that lead to particularfire and fire hazard contexts; to integrate stakeholdersinto all phases of the project from design throughimplementation; and to make the final product avail-able through the Web, using advanced map servertechnology. The model will provide spatially explicit,fine-scale data layers, which in turn will allow users toproduce fire hazard maps, based on different scenarios,for each of the four specific study areas. The model isbeing designed to provide the following benefits:

Figure 2. Schematic diagram of the factors used in the integrated model, including socioeconomic factors, as well as landuse and climate factors.

• Improved ecosystem management• Improved capacity to reduce fire threat, plan

for prescribed burns, and manage watersheds• Better integration of climate and human fac-

tors information into wildfire hazard assess-ments, at longer time scales and at a variety ofspatial scales

• Availability of detailed fire history data andmaps

• Potential for improved policy making at theurban-wildland interface

• Improved capacity to engage in effective pub-

Human DimensionsModel Inputs

Natural and PhysicalProcesses Model Inputs

Decision

IntegratedModel

MitigatingFactors

Assessment(Impact and Risk)

VegetationGreeness

Fuel LoadTopography

Veg & FireHistory

Land Cover

SoilCharacteristics

Moisture(Soil & Veg)

Legislation

SocialFactors

Jurisdiction

HistoricalPolicy

Influences

ENSO

Precipitation& Temp

PDO

AtmosphericCharacteristics

Other

PopulationDistribution

Access Land Use

PopulationDensity

EconomicFactors

Infrastructure

Real EstateValue

Other

44

Figure 1. CEFA home web page (http://www.dri.edu/CEFA).

lic relations and education about wildfire andfire management

• Enhanced capacity to manage for carbon se-questration

The products will include an ArcInfo-based GISmodel, including three-dimensional models of thestudy areas; fire hazard maps and data layers; support-ing data, graphs, and tables; user instructions; a finalreport; and peer-reviewed research publications in ap-propriate professional and scientific journals. A specialsymposium will be held at the end of the project; thissymposium will include a workshop for participants toexperiment with using the model.

Related ResourcesCLIMAS Fire and Climate in the Southwest WebPages http://www.ispe.arizona.edu/climas/fire/

Program for Climate, Ecosystem and

Fire Applications Overview

Tim Brown (Desert Research Institute)February 16, 2001 9:00 AMMarch 28, 2001 10:50 AM

What is CEFA?CEFA, the Desert Research Institute (DRI) Programfor Climate, Ecosystem and Fire Applications (CEFA),was formed on October 1, 1998 through an assistanceagreement between the Bureau of Land Management(BLM) Nevada State Office and DRI. As of November2000, a new 5-year assistance agreement was signedwith the BLM national Office of Fire and Aviation atthe National Interagency Fire Center to perform basicclimate studies and product development for fire man-agement at the national level. CEFA resides within theDivision of Atmospheric Sciences of DRI at Reno, Ne-vada, and works closely with the Western Regional Cli-mate Center (WRCC).

The primary functions of CEFA are to:

• Perform studies and applied research to im-prove the understanding of relationships be-tween climate, fire and natural resources.

• Serve as a liaison between the user and the re-search community by providing product train-ing, assisting in technology transfer and elicit-ing user feedback.

• Provide climate and weather information di-rectly for fire and ecosystem decision-makingand planning.

• Improve operational fire weather forecastingusing new knowledge of climate and meteo-rology.

• Develop and maintain a data warehouse forfire, ecosystem and related climate information.

• Develop tools for fire applications.• Provide a social interaction component related

to climate and wildfire.

CEFA partners include:

• Bureau of Land Management• U.S. Forest Service• California Department of Forestry and Fire

Protection• National Park Service• California Interagency Fire and Forecast

Warning Units• Scripps Institution of Oceanography• California Applications Program• Experimental Climate Prediction Center• Western Regional Climate Center

CEFA maintains a website (http://www.dri.edu/CEFA;Figure 1) that was developed and is maintained to alarge extent from input by users in the wildfire deci-sion-making and planning community. Many ofCEFA’s products are the result of project suggestionsfrom the user community including those from lastyear’s fire-climate meeting here in Tucson. I suspectthat we will engage in several projects as a result of theoutcome of discussions from this meeting.

45


CEFA Research and ProductsCEFA has been performing research and developingclimate products of relevance to wildfire management.A few of these projects are given here along with ex-ample products, but many additional ones are also inprogress or are being initiated.

Recently CEFA developed a lightning climatology forthe entire western U.S. using data from the AutomatedLightning Detection System (ALDS) operated by theBureau of Land Management from April 1985through November 1996 (Figure 2). We are currentlyin the process of developing a new lightning (cloud-to-ground strike) climatology over the western U.S. usingNational Lightning Detection Network™ (NLDN;1990-present) data from Global Atmospherics, Inc. In-dividual strike counts will be binned into a 0.5° spatialresolution grid across the western U.S. for 1990-2000.Maps of monthly strike counts and hour by monthcounts will be produced and made available to agencyusers on the CEFA web site. Since NLDN data areproprietary, access to these maps will be limited to gov-ernment wildfire agencies that pay annual accesscharges for these data.

CEFA has developed a surface climatology of fireweather variables for the Great Basin including tem-perature, relative humidity, wind speed and wind di-rection from Remote Automatic Weather Stations(RAWS). These products are similar to lightning inthat monthly and hour by month climatologies areavailable in graphics and animations. We have also re-cently performed a network analysis of Great BasinRAWS sites. One of the primary purposes of this studywas to assess the adequacy of the network for fire dan-ger and other purposes. Figure 3 shows RAWS loca-tions in Nevada (circle symbols) with a background ofthe Köppen climate classification provided by theIdaho State Climate Services office. CEFA is alsoworking, in collaboration with the California WildfireAgencies and WRCC to perform quality control onhistorical RAWS data and improved metadata for ap-proximately 240 RAWS sites in California. These im-proved data will then be available to perform fire dan-ger rating analyses and determine RAWS climatologi-cal characteristics across the state.

One major project currently in progress is the climateanalysis of the 2000 season. This project is an analysisof the most relevant climatological factors responsiblefor the year’s activity across the western U.S. The pri-mary project goal is to develop an understanding of

Figure 3. Nevada Remote Automatic Weather Station(RAWS) locations (circle symbols) and Köppen climateclassification provided by the Idaho State Climate Servicesoffice.

Figure 2. Automated Lightning Detection System (ALDS)August lightning strike climatology (1985-1996 period) forNevada.

46

Figure 4. An example of CEFA experimental daily NCEPEta model smoke management forecasts for Californiaand Nevada.

Figure 5. Example hourly fire danger map for California.

the climate patterns preceding and during the fire sea-son. Specific tasks include (1) analyzing climateanomalies on monthly and seasonal time scales for anumber of surface and upper-air variables (e.g., tem-perature, precipitation, relative humidity, wind speedand direction, lightning); (2) relating this informationto National Fire Danger Rating System (NFDRS) in-dices such as 1000-hour fuel moisture; and (3) assessthe possible role of La Niña or other teleconnectionpatterns in relation to the fire season.

CEFA’s current primary operational product is experi-mental daily NCEP Eta model smoke managementforecasts (Figure 4). This experimental product consistsof forecasts of mixing height and mean transport windspeed and direction from the most recent 00 UTC and12 UTC Eta model run for the western United States.Graphic forecast displays are available for 6-hour peri-ods out to 48 hours. The California Interagency Fireand Forecast Warning Units receive forecast data in atext format. The project is a collaboration betweenCEFA and the California Wildfire Agencies, partly inresponse to the need for air quality and smoke man-agement forecasts in California.

CEFA is currently in the process of developing a pro-totype operational hourly fire danger map for Califor-nia (Figure 5). It has been noted that in many areasacross the state fire danger can change substantiallyduring a 24-hour period. In an attempt to identifythose situations where fire danger rapidly rises or de-creases during the late night or early morning hours,we are producing hourly maps for each fire danger rat-ing area across the state using components of theNFDRS and hourly RAWS data.

CEFA, working with the Scripps Institution of Ocean-ography joint NOAA Office of Global Program Cali-fornia Applications Program, is taking part in the De-partment of Energy’s Accelerated Climate PredictionInitiative (ACPI). Our primary objective is to examinefuture potential change and variability in seasonal firedanger based on NFDRS indices such as the energy re-lease component and burning index. The analysis willbe done using approximately 150 years (1950-2099) of6-hourly climate model output for a large portion ofNorth America. Elements such as temperature, relativehumidity and precipitation available from the NationalCenters for Environmental Prediction (NCEP) re-analysis and the National Center for Atmospheric Re-search parallel climate model are used as input forNFDRS calculations. The results will allow us to ex-amine regional changes in fire danger strictly as a func-tion of climate variability and trend, and can be usedfor economics and land use policies related to futuresuppression and rehabilitation planning.

CEFA also makes available on our web site productssuch as:

• Standardized Precipitation Index (SPI) maps.• A number of monthly and seasonal weather

and climate forecasts.

47


• A variety of operational and assessment prod-ucts for fire and climate monitoring

Most of these products are produced by other agenciesand organizations, such as WRCC, Scripps Institutionof Oceanography Experimental Climate PredictionCenter, and the International Research Institute. Forsome of these products we provide value-added infor-mation. A primary purpose here is to provide one-stop“shopping” access to climate products and informationrelevant to wildfire management.

Climate, Fire, and the Need for a

National Climate Service

Jonathan Overpeck (Institute for the Study of PlanetEarth, The University of Arizona)February 16, 2001 9:30 AM

Introduction: Why a National Climate Service?We have worked out the dynamics of the El Niño-Southern Oscillation climate process, but these achieve-ments failed to excite interests in Washington DC. Thenational climate community is now trying to figure outwhere to get the resources to move forward. Climateprediction, facilitated by the wealth of data being pro-duced by terrestrial, marine, and atmospheric observingsystems, is moving forward. But the state of knowledgeis not yet as good as it needs to be. We need a global cli-mate observing system that includes institutionalmechanisms as well as satellites and other observinghardware. To achieve this synergy, we need to work fromthe bottom up, while at the same time entities at thefederal level continue to work from the top down.

Regional decisions require integration of climateknowledge into a context that recognizes the existenceof multiple stresses and that recognizes the specific at-tributes and challenges of the places where people liveand work. There is a new initiative afoot to introduce aregionally based, national climate service, analogous butnot identical to the National Weather Service. Thedriving force behind the initiative is emanating fromthe fire community, as well as from stakeholders andjurisdictions at all levels from the local to the interna-tional. Thus, the goal is to provide useful informationto the wide array of stakeholder sectors, while at thesame time acknowledging that the concerns of differ-ent stakeholder sectors may sometimes conflict.

We have learned, through our research and interac-tions with stakeholders—including fire managers anddecision makers—that there are many commonalitieswith regard to areas of risk and information needs. In-deed, demand is growing for place-based science. Butto respond to this demand, we must convince our con-gressional representatives to be constituents for thistype of science. Developing and maintaining stake-holder partnerships is fundamental, as is developingand maintaining mutual trust; this can only take placeat local to regional scales. Likewise, regional partner-ships must be sustained in order to sustain credibility.Such relationships must be ongoing; they cannot beterminated at the end of any given research phase orspecifically funded research project.

A well-designed climate services operation is needed toensure that this occurs. Our work must be built on thepremises of sustained responsiveness (which requirescontinual evaluation), continually improving commu-nications (including honesty with regard to what canand cannot be delivered to stakeholders, given currentscientific knowledge and technological capacity), andever improving science. At the same time, major gapsin scientific knowledge must be addressed at the re-gional level.

For example, in the Southwest, topographic complex-ity poses challenges to monitoring, understanding andmodeling regional climate variability. Among the pro-cesses that are high on the research agenda for theSouthwest are monsoon dynamics and forecasting,linkages between climate-snow-hydrology, and fore-casting wind, relative humidity and atmospheric stabil-ity. Equally important is developing an understandingof climate impacts on human and natural systems thatwill allow development of integrated decision supportsystems.

A Model for Building a NCS and Some Things toPonder…Bottom-up advocacy of science oriented toward usersat the regional level is essential. This involves threesteps:

1. Identification of the knowledge/informationneeds of stakeholders

2. Advocacy of existing programs that are meet-ing regional needs

3. Advocacy for new efforts to fill the gaps

An optimal timeline for development of a formal, re-gionally-based and regionally-focused climate services

48

operation would progress from an initial proportion of90% science and 10% operations, to 50% research and50% operations by 2025. To make this happen, thereneeds to be (1) a more effective consortium amongagencies, universities, NOAA and stakeholders, (2) aregional project management office, (3) a regionalsteering committee, and (4) NOAA-led national ac-tivities.

Many questions remain to be answered as the initiativeunfolds; for example:

• How broad should the climate service mandate be?Should it be “environmental services” instead of amore narrowly focused “climate services”? If this isthe case, could such an operation form the basisfor a new level of bottom-up interagency coopera-tion and partnership?

• Would one-stop shopping and a single voice servestakeholders better?

• Would this lead to broader support for scienceand/or federal land managers’ missions, objectives,goals?

• Before developing new climate products and ser-vices, how can we use what is already available? Towhat extent has this already been tried?

• What is the current political reality? What is themost “sellable” package in today’s political world?

- We need to stress the multiagency aspects.- We need to work with other states.

Moreover, there are other points to keep in mind:

• No scale matters more than any other.

• We need more local interaction; face-to-faceinteractions are important.

• The research and outreach activities we arecarrying out now, particularly through in-volvement of students and community mem-bers, are allowing us to produce essential next-generation knowledge.

• NOAA should be funded at the billion dollarlevel to achieve the kinds of goals outlinedhere.

• We need to make it clear to Congress howmuch it really costs to “put out a fire.”

• We need to take seriously concerns about con-flicting responsibilities and accountability.

• We need to assure that the regional climateservices initiative does not get absorbed intothe global warming debate.

• We need to continue supporting improve-ments and enhancements to our observationsystems.

• We need to continue working to assure thatthe work and insights of the regional assess-ments are well integrated into the climate ser-vice framework—at the national level, climatefactors and impacts are beginning to be re-flected in planning documents; what is miss-ing is that some agencies are not thinking hardenough.

• We need to continue working to raise congres-sional awareness of the issues.

49


Fire and Climate in the Southwest 2001

Climate of the Southwest

Andrew Comrie (Department of Geography and Re-gional Development, The University of Arizona)March 28, 2001 9:00 AM

Basic Characteristics of the SW ClimateClimate v. Weather. Weather includes events that occurover timescales of minutes to several days, whereas cli-mate is, roughly speaking, a summary of weather overtimescales that range from months to thousands ofyears. However, climate is more than just the averageweather. It is the range of atmospheric conditions at alocation experienced or expected over time. It is im-portant to note that climate is always changing. Whatwe think of as normal climate is a relative concept.Even though it is convenient to use the convention ofthe average of the most recent 30 years’ weather as“normal” climate, we know that climate changes overthe course of a century, and that records of changesover several centuries show 30-year periods that wouldseem quite far from what we presently consider to benormal. To use a financial market analogy, weather islike day trading, whereas climate is like long-term in-vesting. In long-term investing, large forces (climate)shape the general performance of companies over time,but they play out as complex daily patterns (weather)of ups and downs on the market.

Defining the Southwest Climate. For our purposes thecore region of SW climate is AZ and NM, with someextension into northern Mexico and the Upper Colo-rado River Basin (Figure 1). The annual average pre-cipitation totals shown in figure 1 indicate spatialvariation of precipitation over the Southwest region.The complex topography of our Southwest regionleads to intricate spatial patterns of precipitation. Pre-cipitation increases with elevation, as seen in the hightotals over the Mogollon Rim, sky islands, SouthernRockies and Colorado Plateau. However, precipitationis sparse in regions in the lee of mountain ranges, rela-tively far from moisture sources, and those out of thepath of the dominant westerly flow in the NorthernHemisphere midlatitudes. Thus, there are very lowprecipitation amounts in northeastern AZ and north-western NM. Overlaid on figure 1 are the boundariesof the NOAA state climate divisions, which are re-

gional averages of NOAA-NWS cooperative observa-tion network stations.

Basic Characteristics of the Southwest Climate. Figure 2shows average monthly precipitation totals for each ofthe climate divisions in our region. Seasonal precipita-tion varies throughout the region, but for the mostpart in AZ and NM there is a primary summer pre-cipitation maximum, accounting for approximately50% of annual precipitation, and a secondary maxi-mum in winter, accounting for approximately 30% ofannual precipitation. Summer and winter are, however,largely unconnected, and are governed by altogetherdifferent climatic processes (see below). Temperaturefollows the typical seasonal cycle, with maximumtemperatures during the summer months and mini-mum temperatures during winter months. Averagetemperature in our region is strongly influenced byelevation.

Atmospheric Controls of SW ClimateSome of the key characteristics of our Southwest semi-arid/desert climate are the low precipitation, clear skiesand warm weather that attract so many people to our

Figure 1. Average annual precipitation totals for theSouthwest core region. NOAA state climate divisionboundaries are shown. Figure from Sheppard et al. 2001,“The climate of the Southwest,” in review for thepublication, Climate Research.

50

region. These characteristics are the result of severalfactors, chief of which is the subtropical high-pres-sure that dominates our weather from much of theyear. The semi-permanent high-pressure cells over theNorth Pacific and North Atlantic oceans are the re-sult of the return flow from thermally direct atmo-spheric circulation originating in the tropics. Thesubsiding air in these high-pressure cells inhibits pre-cipitation.

Topography introduces spatial variation in precipita-tion and temperature; areas in the lee of mountainranges experience the so-called rainshadow effect, inwhich much of the rainfall is forced out of moist airparcels as they rise on the windward side of highmountain ranges. Moreover, proximity to moist airplays a key role in the spatial distribution of precipita-tion. The moisture sources for our region include theEastern Pacific, which dominates in winter, and theGulfs of Mexico and California, which provide mois-ture for our summer rainfall. In general, the aridity ofour region is accompanied by high temperatures andhigh rates of evapotranspiration.

The climate of our region is strongly affected by feed-backs between the atmosphere and the ocean. Thus, ElNiño and La Niña, which result from the effects of Pa-cific Ocean heating and cooling, respectively, on theatmosphere, play a key role. The large degree of tropi-cal Pacific Ocean heating and cooling during El Niñoand La Niña, affects atmospheric moisture availability,as well as the path of storms over distant locations,such as our region. Such long distance connections inatmospheric phenomena are called “teleconnections.”It is important to note that although on average ElNiño enhances winter precipitation in our region andLa Niña results in dry winters, no two El Niño or LaNiña events are exactly alike.

Wintertime Atmospheric Controls. Figure 3 shows typi-cal winter storm tracks over the U.S. The midtropo-spheric jet stream, which steers high and low-pressuresystems, typically enters the western U.S. at around theU.S.-Canada border, as shown by the red line. Notethat the storm track follows a sinusoidal path, as it isforced by the Rocky Mountains to ridge northwardand trough southward once over the Eastern U.S. Thistypical storm track results in dry winters in the South-west; this effect is enhanced during La Niña. Wet win-ters frequently result from an amplification of the ridg-ing and troughing, which allows more moist tropicalair to enter our region. This amplified pattern is fre-quently referred to as the Pacific/North American(PNA) pattern, or “meridional flow.” El Niño can leadto this kind of enhancement of the PNA pattern, or itcan lead to split flow, in which the dominant westerlywinds split around high-pressure in the Eastern Pacific,and storms are steered south into our region.

Summer Atmospheric Controls. The dominant westerlyflow that characterizes winter atmospheric circulation

Figure 2. Seasonal precipitation cycle for NOAA climatedivisions as shown by graphs of average monthlyprecipitation for each climate region. Note the increase insummer precipitation peak in southeastern AZ and acrossNM. Figure from Sheppard et al. 2001, “The climate of theSouthwest,” in review for the publication, ClimateResearch.

Figure 3. Average midtropospheric (700 mb) storm tracks.Typical flow results in dry winters in the Southwest,whereas meridional or split flow often associated with ElNiño results in wet winters. Figure from Sheppard et al.2001, “The climate of the Southwest,” in review for thepublication, Climate Research.

51


across the western U.S. moves northward during thesummer in response to atmospheric heating in theNorthern Hemisphere. Heating over the North Ameri-can landmass draws warm moist air in from adjacentocean regions to the southeast and southwest of thecontinent. A seasonal peak in rain accompanies thisseasonal change in wind direction from July throughearly September in the Southwest. The overall circula-tion is called the North American Monsoon and it ispart of a larger continental-scale circulation patternover the Mexican highlands. The moisture sources arethe Gulfs of California and Mexico, as well as the east-ern Pacific Ocean. The effects of El Niño and La Niñaon the monsoon are unclear, due to complex interac-tions between moisture characteristics over the land,and sea surface temperatures over the oceans. Some-times, but not always, dry winters (La Niña) precedewetter summers. The monsoon varies within the sum-mer season in the form of “bursts” and “breaks,” thatis, periods when storms are active or inactive for 1-3weeks at a time; these bursts and breaks are connectedwith surges of moisture from the Gulf of California.Also in summertime, tropical hurricanes are a factor,especially during late summer/early fall. In general,they are relatively rare in the SW, though they cancause substantial rainfall events.

Climate Variability over Time & SpaceClimate Change Over Time. The instrumental climaterecord extends back only about 100 years. Thus, weneed to use paleoclimate records in order to extend ourknowledge of climate variation back in time. Due toan abundance of long, well preserved, tree-ringrecords, we are able to extend Southwest climaterecords back in time 1000 years or more.

The Palmer Drought Severity Index (PDSI), a singlevariable derived from variation in precipitation andtemperature, is well reconstructed from tree-ringrecords. The PDSI is a handy representation of SWprecipitation and, to a lesser extent, temperature. Fig-ure 4 shows variation in moisture as represented by re-constructed PDSI. It is most important to note thatthe Southwest has experienced frequent dry and wetperiods, that vary widely in intensity and timing. Thesustained drought of the 1950s was clearly among theworst in the last 1000 years. Significant wet periodsoccurred around 1726, 1793, 1839, 1868, and 1907.The greatest annual wet-dry switch was from 1747 to1748. If we take a long-term view of SW moisturevariability, we can see that the 20-year wet period ofthe early 1900s was exceeded only in the early 1600s,

and the longest drought of the millennium was in the1500s. The smooth red line in figure 3 shows low-fre-quency moisture variation, and exhibits a ~80 year pat-tern of variation characterized by alternating below toabove average PDSI. This feature might possibly relateto fluctuations in solar output. The recent rise in tem-perature (figure not shown) is unprecedented in thelast 400 years. Temperature records show a ~20-yearpattern of variability, with significant warmth duringthe mid-1600s and 1930s, and cold periods commenc-ing around 1907 and 1600. The greatest extremes inthe temperature record are relatively recent, with ex-treme warmth in 1865, 1881, 1934, (equaled around1651) and cool in 1725, 1835, 1866 and 1965.

Recent work from the CLIMAS project documentsspatial variability in SW climate. We are in the processof gridding temperature, precipitation and PDSI at 1km resolution, with an improved algorithm to capturevariation due to elevation, slope and aspect.

ReferencesSheppard, P., A. Comrie, G. Packin, K. Angersbach,and M. Hughes, 2001. The Climate of the Southwest.

Figure 4. Reconstructed Southwest PDSI for the pastmillennium (top) and the past 300 years (bottom). PDSI isa drought index of moisture variability. The blue lineindicates interannual variation; the red line indicates low-frequency variation; the green line indicates instrumentalPDSI values. Years of unusually high or low reconstructedPDSI are noted in the bottom graph. Figure fromSheppard et al. 2001, “The climate of the Southwest,” inreview for the publication, Climate Research.

52

Climate and the 2000 Fire Season in

the Southwest

Tom Swetnam (Laboratory of Tree-Ring Research,University of Arizona)March 28, 2001 9:40 AM

Background on the Relationship Between Climateand FireRegional to continental-scale fire years are a naturalfeature of the western US. These exceptional fire yearsare usually associated with droughts that begin in orpersist through the winter preceding the fire year. Insome regions they are also related to wet conditions inprior seasons and years. Lags in fuels and climate con-ditions associated with exceptional fire years providean opportunity for early warning. The 2000 fire seasonis a classic case in point.

The largest fire year in the United States in the 20th

century was 1910, a year similar to 2000 in that largefires occurred in almost every western state. More than5 million acres burned across the western US, of whichmore than 300,000 acres were located in the South-west. Large fires typically account for more than 95%of all area burned through time. A key aspect of suchregional fire years and large fire events is that they re-sult in highly synchronized fire activity at regional tocontinental scales. The 2000 fire season occurred inthe historical context of an increasing trend in acresburned over much of North America (Figure 1).

Using Tree-Rings to Record Fire HistoryFires that do not kill a tree often leave a scar (catface),which is recorded in the tree’s annual growth ring (Fig-ure 2). By carefully examining the tree rings, we candetermine the year and season in which the fire oc-curred. Tree-ring core and stump sections are obtainedfrom 20 to 30 different trees per acre in each studysite; the collection of multiple samples is necessary be-cause no individual tree gives a complete history. Spa-tial scales of tree-ring analysis range from that of indi-

vidual stands to the watershed level, entire mountainranges, up to the broader regional scale. At this largerscale, broad patterns can be discerned throughcrossdating of tree rings. This allows identification ofthe year a fire occurred and the synchrony of fire acrossspace. We have 300-500 years of fire history data inthe Southwest, based on more than 100 sites and sev-eral thousand fire-scarred trees (Figure 3).

Figure 1. Area burned appears to increase in many areasof North America since the 1960s. Possible causes ofthese trends include fuel accumulations since effective firecontrol began, and climatic change (Source: Swetnam T.and J. Betancourt, 1998: Mesoscale disturbance andecological response to decadal climatic variability in theAmerican Southwest. Journal of Climate 11:3128-3147.)

Figure 2. (A) Tree in the process of being fire scarred. (B)Dated fire-scarred tree-ring slab specimen. Photoscourtesy of UA Laboratory of Tree-Ring Research.

1920 1930 1940 1950 1960 1970 1980 1990

01020304050607080

1920 1930 1940 1950 1960 1970 1980 1990

02468

1012141618

1920 1930 1940 1950 1960 1970 1980 1990

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Canada

11 Western States

Arizona and New Mexico

In review for Climate Research.

Related ResourcesThe Climate of the Southwest (P. R. Sheppard et al.)http://www.ispe.arizona.edu/climas/reportseries/index.html

53


To examine long-term relationships between climateand fire in the southwestern United States, regionalvalues for tree-ring growth in drought-sensitive treesgrowing at sites unaffected by fire are compiled. Thesevalues can then be used to reconstruct gridded PalmerDrought Severity Index (PDSI) values. The fire historyrecords are then compared with reconstructed PDSI(and other proxy records, including records of ElNiño-Southern Oscillation [ENSO]). These compari-sons highlight the effects of climate on fire occurrence.

Multi-Century Fire-Climate RelationshipsLong-term trends in fire scar records show a drop-offin frequency and extent of fires due to grazing and firesuppression which began around 1900 (Figure 4). An-nual area burned in the 20th century was probably oneor two orders of magnitude lower than in the pre-20th

century period, but a recent increase is evident afterthe 1960s (Figure 1). This increase may be associatedprimarily with a 20th century increase in fuels and for-est density, but climate variability probably played arole as well.

In addition to decadal trends, historical fire climatologyalso illustrates the importance of interannual patterns.In the Southwest, for example, the year 1747 was verywet and had low fire activity, and 1748 was the mostsynchronous (extensive) fire year in the past 300 years(Figure 4). Historical reconstructions of ENSO indicatethat 1747 was probably an El Niño Year and 1748 a LaNiña year. Similar patterns of rapid switching of wet todry conditions have been identified in assessments of thepast three centuries of fire and climate in the Southwestand in the Colorado Front Range.

These combinations of wet and dry years are probablymore important at lower elevations, where presence offine fuels (grasses, needles, etc.) is a limiting factor forfire ignition and spread. In semi-arid landscapes these“fine fuels require 2 to 5 years, or longer to recover fol-lowing a surface burn. Most conspicuously, the wet/dry pattern is often associated with ENSO, when dryLa Niña years follow on the heels of wet El Niño years(Figure 5). Fortunately, our ability to forecast climaticconditions, especially with regard to ENSO, hasgreatly improved.

The Cerro Grande FireIn the Bandelier National Monument and the Santa FeNational Forest area, the high elevations (~10,000 feet)are typically characterized by fir and spruce forests andmontane grasslands. At lower elevations, ponderosa

Figure 3. Southwest tree-ring fire history chronology sites.(Source: UA Laboratory of Tree-Ring Research.)

Figure 4. Synchrony of fires in the Southwest U.S. Notethe decrease in number of sites recorded fires due tograzing and fire suppression activities beginning at theend of the 19th-century. 1748 is the most synchronousyear in the record, a classic example of large-scale firewhen exceedingly dry years follow wet years.

pine dominates. The Cerro Grande prescribed fire inMay 2000 was set in the high elevation environment,where conifers had invaded former montane grasslands.

The fire prescription called for fairly dry conditions forthis burn to succeed in consuming fuels and killingsome trees, which was needed to restore the montanegrassland.. Sufficiently dry conditions may occur athigh elevations only during moderate to extremely dryyears. Given these kinds of relatively dry conditionsneeded for prescribed burning at high elevations, extrapreparedness and precautions are needed to avoid thepossibility of fire reaching lower-elevations, wheremore extreme fire behavior can be expected. Prescribed

54

burning in dry springs has always been hazardous inthe Southwest, especially because high winds can occurwithout much warning (see Snook, this volume).There were undoubtedly some factors that led to theCerro Grande Fire disaster which could not have beenanticipated in advance. However, it is my opinion thatgreater awareness and incorporation of long-term andbroad-scale fire history and climatological patternscould assist in avoiding future disasters of this kind.

Into The FutureIn my view, we need to do more to plan for worst-casescenarios. The potential hazard of escaping fires associ-ated with prescribed burning during drought years,e.g., extreme La Niña years following wet years, re-quires fire managers to exercise a higher level of cau-tion during dry springs that follow dry winters. Wealso must recognize that, although long-range climateforecasts have a high degree of uncertainty, this doesnot mean that the “misses” during some years (e.g.,1999 was forecast to be a very bad season in the South-west, but was only average) mean that future forecastscan be safely discounted. An analogy is hurricane fore-casts. Forecasts of hurricane landfalls are frequentlyused to warn communities for possible evacuation, andoften the hurricane veers off the forecasted pathway atthe last minute. It would be foolhardy to use such er-ror in past forecasting as justification to discount fu-ture forecasts.

Our planning and preparations for exceptional fire sea-sons do not yet fully incorporate our understanding ofmulti-year and multi-decade climate changes. If, forexample, we are shifting into an extended period ofwarmer sea surface temperatures in the southeasternNorth Pacific, then there is a higher likelihood of ex-tended drought in the Southwest. I recommend build-ing a larger fire fighting resource base, one that mightseem excessive in average fire years, but which wouldbe available during the big fire years. Such a strategywould probably turn out to be more cost-effective thanthe current process of raising and allocating extra re-sources on short turnaround during high fire years. Inaddition, better climate information, such as the cli-matology of wind, would help put potential fire hazardin a better context.

Finally, I suggest that we all exercise some humility inthinking about fire. There is much that we do notknow, especially with regard to the climatic and meteo-rological mechanisms and processes that drive large fireevents and regional fire years.

Figure 5. Wet conditions in the Southwest during the1997-1998 El Niño increased fine fuel loads andunderstory growth. Dry conditions were brought on by thelate 1998-2001 La Niña. The combination of wet and dryyears created conditions right for large-scale fire. (Source:NOAA Climate Prediction Center).

Related ResourcesTom Swetnam’s Dendroecology Reprints (PDF format)http://tree.ltrr.arizona.edu/~tswetnam/pdf.htm

CLIMAS Fire History Analysis Using Tree Ringshttp://www.ispe.arizona.edu/climas/fire/overview/history.html

55


Fire and Emergency Management in

Arizona

Alex McCord (Arizona Department of EmergencyManagement)March 28, 2001 10:10 AM

The purpose of this very brief talk is to give somebackground about the Arizona Division of EmergencyManagement’s fire management activities. Arizona Di-vision of Emergency Management (ADEM) coordi-nates emergency services and the efforts of governmen-tal agencies to reduce the impact of disasters on per-sons and property. We do this through (a) mitigationactivities, which may lessen the damages and lossescaused by various hazards; (b) preparedness activities,such as planning and training, which enable the stateand local governments to better respond to emergencyevents; (c) response activities during actual emergen-cies, such as the issuing of warnings, mobilization ofresponse personnel, coordination of resources; and (d)recovery activities after an emergency or disaster whichassist local governments and individuals to restore pub-lic facilities, homes, and businesses. The objective of allof our programs, upon the occurrence of a fire emer-gency in Arizona, is to minimize injury and loss of life,reduce personal property damage and economic loss,and to restore essential community and public services.

ADEM’s major wildfire concerns are at the urban-wildland interface. We would like to take action tomitigate against any “Los Alamos types of situations.”The public perceives a basic conflict/contradictionwith regard to using fire for forest restoration. Thus,there is resistance to prescribed fires. We need a betterway to deal with these kinds of issues at the urban-wildland interface. In particular, we would like to ini-tiate a community education program, stressing thingslike the need to clear around residences, the hazard ofshingle roofs, etc. However, we lack the funding forthis. Along these lines, a few years ago we began a di-saster-proofing and education initiative called ProjectImpact, oriented around Tempe and Yuma as pilotcommunities. We held a workshop in order to bringthe Project Impact communities together to discusstheir successes, challenges and projects. We still awaitfunding to continue this project and urban-wildlandcorridor work in Yavapai County.

One ray of hope for fire emergency management inArizona is FEMA, which has tightened rules for fund-ing reconstruction after hazardous events. FEMA has

implemented a rule that they will only pay out once tothose without insurance coverage. This will certainlyhave an influence on urban-wildland interface impacts.Another important concern for ADEM is preparednessfor evacuations. Last year’s large and intense fires madepeople in Yavapai County very nervous. This is espe-cially worrisome for situations, such as church camps,where large numbers of people are dropped off at re-mote cabins and left without any transportation for aweek or more. For example, there can be 5,000 to10,000 people in remote areas of the Bradshaw Moun-tains of central Arizona at any given time during thefire season, with no transportation out in an emer-gency. ADEM would like better climate and fire-relatedproducts that can assist in persuading the Arizona legis-lature to fund mitigation activities in areas such as this.

Related ResourcesArizona Division of Emergency Managementhttp://www.dem.state.az.us/

A Brief Introduction to Climate

Forecast Resources for Fire

Management

Ron Melcher (Arizona State Land Department)March 28, 2001 10:20 AM

In this talk, I would like to introduce you to some cli-mate and fire forecast products (available on the WorldWide Web) used by Arizona State Lands, and to dis-cuss some issues about the correct interpretation ofthese products.

Probably the best of the lot are short-term (1 to 3days) fire weather forecasts. Similarly, short-termweather forecasts from NOAA and local NationalWeather Service offices are excellent for short-term firemanagement. NOAA also provides a series of longer-term outlooks (available at http://www.cpc.noaa.gov/products/forecasts/), which have a variety of virtuesand a variety of problems. Their 6-to-10 day and 8-to -14 day outlooks seem less dependable than most oftheir other products. These outlooks are just plain hardto understand and interpret (Figure 1).

NOAA’s Climate Prediction Center (CPC) also putsout monthly and seasonal climate forecasts (http://www.cpc.noaa.gov/products/predictions/30day/).These forecasts come in a couple of different formats

56

(Figure 2 and Figure 3). The key point with thesemaps is: How do you interpret them? I find the colormaps very misleading. These are not maps of forecastedtemperature and precipitation, they are maps of thechange in probability from normal conditions. How doyou interpret the statewide areas with “CL” in them? Ifyou look carefully at Figure 3, which I find to be mucheasier to interpret, you can see that CL means insuffi-cient skill...in other words they aren’t even making a pre-diction for huge portions of the country. When usingthe color maps, it is really important to check againstthe legend, which isn’t even available on the same page,otherwise it’s all too easy to make the mistake of sayingthat this area is going to have above-average precipita-tion and this area is going to have normal precipitation.

Other products that are useful for monitoring condi-tions associated with fire are maps from the WildlandFire Assessment System (WFAS; http://www.fs.fed.us/land/wfas/welcome.htm) in Missoula. WFAS has mapsthat show greenness, fuel moisture, fire potential in-dex, and fire danger (Figure 4). They also have handyfour-panel displays with forecast climate parameters(Figure 5) and measures of greenness and fuel moisture(http://www.fs.fed.us/land/wfas/4pannd.gif ).

NESDIS (National Environmental Satellite, Data, andInformation Service) also has greenness-type maps offire risk and vegetative health. Their vegetative healthmap is very useful, because it compares a day in thecurrent year with the same day the year before (http://orbit-net.nesdis.noaa.gov/crad/sat/surf/vci/usafr.html).I find this kind of comparison very useful. It can alsoact as a way to gauge current conditions against condi-tions with which I am familiar.

Figure 1. NOAA Climate Prediction Center 6-10 dayoutlook. Note how complicated this figure is and howdifficult to interpret.

Figure 4. Wildland Fire Assessment System daily firedanger map.

Figure 3. NOAA monthly climate outlook for April, 2001.Compare this format with the one in Figure 2.

Figure 2. NOAA monthly/seasonal climate outlook (colorversion). Note that areas marked “CL” lack of a forecast.

57


Fire History in the Jemez Mountains:

The Bandelier Perspective

L. Dean Clark (U.S. National Park Service, BandelierNational Monument)March 28, 2001 1:10 PM

BackgroundThe Los Alamos area gets 9,300 to 24,000 lightningstrikes per year; at Los Alamos, there about 62 thun-derstorm days a year. May and June lightning strikesare important for fire incidence.

From 1800 to 1900, data indicate a close correspon-dence between climate and fire. Since 1900, the corre-spondence has decreased significantly (see, e.g., datacollected along Los Alamos Pipeline Road). Frequentfires ceased when intensive livestock grazing (notablysheep) began, around 1900. Since that time, there havebeen conditions favorable for tree growth and forestexpansion, including fire suppression policy. We knowthat grasslands are sustained through frequent surfacefires. However, with suppression, the ecological dy-namics have changed to favor trees over grass. A cli-mate shift, which brought on extended drought duringthe 1950s, resulted in the movement of ponderosapines a quarter of a mile up the mountain and the

death of ponderosa pine stands at lower elevations.

Prescribed Burns and the 2000 Cerro Grande FireIn order to put the Cerro Grande burn in context, weneed to remember that two large wildfires occurredpreviously in or near Bandelier: the 1977 La Mesa fireand the 1996 Dome fire. The La Mesa fire was ignitedby lightning, whereas the Dome fire was human-caused. The Dome fire was a big fire that got the at-tention of federal entities and Los Alamos. After theDome fire, fire breaks were established at Bandelier.Thus, there was a well established need for an activeprescribed fire program at Bandelier.

Perhaps it is most important to emphasize that the pre-scribed burn program had been successful for 20 yearsprior to the Cerro Grande fire. Photographic evidenceindicates that there was a change in vegetation in thisarea from grassland to ponderosa pine after the 1950sdrought. The trees had taken over about three-quartersof the high elevation grassland. In the early 1990s, for-est managers at Bandelier tried to clear the areathrough light burns, but these did not succeed inchanging the forest structure. All they got were light,patchy, scabby burns. Thus objectives were notachieved. Moreover, the timing of prescribed burns atBandelier is an issue, because burns cannot necessarily

Figure 5. Wildland Fire Assessment System daily fire weather observations and next day forecasts.

58

be scheduled to correspond to times of the year whenfire would be most beneficial to ecosystem dynamics.Natural fires occur during the spring growth period.The situation around the May 2000 fire is that a high-intensity burn was required to achieve the objective ofrestoring the high elevation grassland. However, lower-elevation areas, with vegetation such as aspen andbrush, where no fuel load modification had previouslybeen done, and which are prone to high-intensityburns, were not included in the burn prescription.Wind, a large factor in any burn situation, pushed theinitial fire beyond the boundary of the prescription.

I expect the final report on the Cerro Grande fire to bereleased very soon. I believe that the strength of thedata will get us over the political furor over the fire,and allow us to re-establish a successful and necessaryprescribed burn program.

Related ResourcesLandscape and Fire Ecology Studies at Bandelier Na-tional Monument and the Jemez Mountainshttp://www.mesc.usgs.gov/projects/landscape-fire-ecology.html

Landscape Changes in the Southwestern United States:Techniques, Long-term Data Sets, and Trendshttp://biology.usgs.gov/luhna/chap9.html

Climate Forecasts Overview for the

Southwest

Thomas Pagano and Holly Hartmann (Departmentof Hydrology and Water Resources, University ofArizona)March 28, 2001 1:40 PM

IntroductionIn this talk, I will introduce official climate forecastsmade by the U.S. government, I will talk about how tointerpret these forecasts, what their performance hasbeen to date, the current forecasts and their implica-tions for this fire season. It is important to distinguishbetween weather and climate. Weather forecasts aremade for periods of 0-3 days; long-range weather fore-casts are made for periods of 4-12 days; and seasonal cli-mate outlooks are made for periods of one season to oneyear. Of course, forecasting is quite an uncertain busi-ness, or as Victor Borge said, “ Forecasting is difficult,especially about the future.”

What Are Climate Forecasts and Who Makes ThemOfficial climate forecasts are called long-lead climateoutlooks and they are made by the NOAA Climate Pre-diction Center (CPC; http://www.cpc.ncep.noaa.gov).These products are made by a team of 20 or more fore-casters, blending six classes of tools with human exper-tise. All other climate forecasts are either experimentalor unofficial. The tools used by climate forecasters in-clude physical models, trends (such as long-termwarming), statistical models (including pattern recog-nition and information based on the state of El Niño-Southern Oscillation [ENSO]), and human expertise(including knowledge of the strengths and weaknessesof various models, and common sense).

An Example of a Simple Statistical ForecastIn this example we take winter (December-February)precipitation at Willcox, Arizona and make forecastsbased on the state of ENSO. First, we classify the ob-servations by taking all winter precipitation observa-tions from Willcox (Figure 1) for the period 1898-1989, and dividing them into upper, middle, andlower thirds (y-axis in Figure 2), and then dividingthem again into three ENSO modes, La Niña, non-Niño and El Niño based on an El Niño sea surfacetemperature (SST) index (x-axis in Figure 2). Next wecount the number of observations in each category.Then we convert the tallied observations to percent-ages. Now we can begin to make forecasts based onour record. We can say, for example, if it is a La Niñayear, then there is a 10% chance of wet conditions,30% chance of normal conditions and a 60% chanceof dry conditions in Willcox, AZ. Note that there isnever a 100% chance of any of these three precipita-tion conditions; however, there is a very poor chanceof wet conditions during La Niña and, conversely, avery poor chance dry conditions during El Niño. Wecan do the same sort of exercise for summer monsoon

Figure 1. The location of the town of Willcox insoutheastern Arizona.

59


precipitation. Summer precipitation, however, showsfar more variability during the different phases ofENSO. Therefore, there is far less predictability, andonly a 5% increased chance of wet conditions duringLa Niña and a 5% increased chance of dry conditionsduring El Niño.

Understanding CPC ForecastsIn order to understand CPC forecasts, it is important toknow what CPC considers normal. CPC uses as itsbaseline the most recent 30 years of record. At the timeof this conference the CPC baseline is 1961-1990; laterthis year the baseline will change to 1971-2000. Theythen select the 10 wettest years, which are consideredwet, the 10 middle years, which are considered normal,and the 10 driest years, which are considered dry. CPCthen expresses their climate outlooks maps in terms ofthe increase in the probability of being in the wet or drycategory. Thus, in the map in Figure 3 there is a 5% in-creased chance of wet conditions in Seattle and a 10%increased chance of dry conditions in Miami. Put an-other way, the probability of precipitation for Seattle is

• 38% chance of wet• 33% chance of normal• 28% chance of dry

and the probability of precipitation for Miami is• 23% chance of wet• 33% chance of normal• 43% chance of dry

In Figure 3 the CL over Boise, Idaho indicates thatforecasters are completely uncertain about what is go-ing to happen. Thus, the probability of precipitationfor Boise is

• 33% chance of wet• 33% chance of normal• 33% chance dry

Note that private sector companies often based theirforecasts on the CPC forecasts, but they substitute nor-mal for CL. This is a very bad misinterpretation of theactual forecast. In fact, it would probably be better ifthe CPC used a designation such as “no skill,” “noforecast,” or “don’t know,” rather than CL.

CPC Forecast PerformanceForecast performance is multifaceted, and some usersrequire information about false alarms (i.e., predictionsthat never came to bear), whereas other users requireinformation about surprises (i.e., situations that cameto bear, but were never forecast). One measure for

Figure 2. A simple statistical forecast based on pastprecipitation at Willcox, Arizona. (a) Willcox winterprecipitation divided into terciles (y-axis; wet, normal, dry)and ENSO phase (x-axis). (b) Tally of the number of timesWillcox winter precipitation falls into each tercile. (c)Percent chance of winter precipitation falling into eachtercile, based on ENSO phase. Note the increasedlikelihood of dry conditions during La Niña.

Figure 3. Official NOAA Climate Prediction Center climateoutlook for March-May, 2001 precipitation.

60

evaluating probabilistic forecasts is the ranked prob-ability score, which rewards strong probability state-ments that fall into the correct category and penalizesprobability statements that are far off the mark. Workby Holly Hartmann, evaluating all CPC seasonal fore-casts for a particular season (Figure 4), shows thatCPC’s forecasts for pre-monsoon precipitation havehad negative skill, whereas forecasts made during thesummer and fall for fall and winter precipitation, re-spectively, have pretty good skill, especially for Ari-zona. CPC temperature outlooks for fall and winterfor the Southwest have very high skill.

The Current ForecastAt present, we have a La Niña that is nearly dead, i.e.,very weak negative SST anomalies in the equatorial Pa-cific. There is a weak tendency toward wet summers inthe Southwest during La Niña. For April, the climateoutlook for the Southwest U.S. indicates the followingprecipitation probabilities: 27% wet, 33% normal,40% dry. As for temperatures, there has been a long-term warming trend in the Southwest. For example,the 1990s were very warm compared to the 1961-1990average, so the most recent climate outlook, which in-corporates this trend as a major predictor, indicates anincreased chance of above-average temperatures for theSouthwest (38-43% chance of warm). Long-term sea-sonal temperature outlooks also indicate increasedchances of above-average temperatures for the South-west, with the highest probabilities centered over west-ern Arizona (Figure 5).

Spring/summer seasonal climate outlooks for precipita-tion (Figure 6) show a marginally lower chance of adry monsoon (27%), chances of a “normal” monsoonare unchanged (33%), and a slightly higher chance of awet monsoon (40%). Increased probability of above-average monsoon precipitation is only for the June-Au-gust and July-September long-range outlooks. There ismuch uncertainty in these outlooks, as we are in atransition in the phase of ENSO.

When no forecast is available, as in the May-July 2001precipitation outlook (Figure 6), it is best to study climatehistory. Learn the range of variability for the region, lookat climate changes for past years, and study the recent pastin order to best discern fuel conditions and possible ef-fects of long-term climate conditions on fuels.

Related ResourcesNOAA/NWS/CPC Suite of Official Forecastshttp://www.cpc.ncep.noaa.gov/products/predictions/

Figure 6. NOAA CPC climate outlooks for summer 2001precipitation (released March 15) call for slightly increasedchances of above-average precipitation for the SW.

Figure 4. CPC forecast performance 1995-2000, based onthe ranked probability score. Winter planning… refers topre-monsoon season forecasts made during winter months;summer planning… refers to post-monsoon seasonforecasts made during summer months, and so on.

Figure 5. NOAA CPC climate outlooks for summer 2001to (released March 15, 2001) call for increased chances ofabove-average temperatures for the SW.

61


Breakout Groups

Breakout Group Recommendations

The participants in each workshop were invited to par-ticipate in large and small group discussions with thegoal of identifying key issues and making recommen-dations for future research, and information transfer,collaboration, and policy. February workshop partici-pants engaged in several discussions, including an openfloor discussion entitled Lessons from the 2000 fire sea-son: What can we do better in 2001? and breakoutgroup discussions on the topic The use of climate infor-mation by fire managers. During the course of openfloor discussions in the February meeting, participantsidentified the following key topics for small group dis-cussion:

• databases and observation networks• information transfer and decision analysis

tools• training

March workshop discussions were more open-ended,however, the topics identified by February meetingparticipants were used as guidelines for discussion.Given that the geographic focus of the February meet-ing was national, whereas the geographic focus of theMarch meeting was the Southwest, the breakout groupreports have been summarized separately, below.

Climate and Fire 2001 (February)

Databases and observation networksBreakout group remarks regarding databases and ob-servation networks tended to fall into three categories,as follows: observation networks, parameters of interest(including issues of spatial and temporal scale), and re-marks synthesizing a number of concerns.

Observation Networks. With regard to observation net-works, the overall sentiment was that we do not andcould never have too much data. Participants sug-gested selective increases in the number of observationstations. There was concern regarding the aging andmaintenance of stations and a well-organized replace-ment program was recommended. One of the keypoints during the workshop was that fire danger in thewestern U.S. is often dependent on winter precipita-

tion; thus, participants recommended an increase inthe number of all-weather observation stations.

Parameters and Research. Participants identified a vari-ety of meteorological and ecosystem variables thatwould aid effective fire management. All breakoutgroups identified wind as a key variable for furtheranalysis. Fire managers expressed interest in greater andeasier access to wind data and summaries of wind data.Participants emphasized to that more wind observa-tions are necessary for good fire management and forfire-climate studies. Participants recommended that ex-treme and locally intense winds, such as Santa Anawinds and Chinooks, be the subjects of further study.The study of the relationship between these extremewinds and large-scale synoptic climate patterns wasrecommended. Participants suggested that variablessuch as precipitation, relative humidity, stability andwind were more important for fire management thanwas temperature. Participants suggested further re-search into climate indicators of fire, such as fire/drought indices (e.g., PDSI, Keetch-Byram), and rec-ommended research into the relationships betweenstandard climate variables, atmospheric circulation andfire/drought indices. Similarly, participants suggestedresearch into the relationships between climate vari-ables and National Fire Danger Rating System data, inorder to provide links between weather, climate, andfire. It was recommended that such analyses incorpo-rate nested spatial resolution at timescales rangingfrom days to seasons, and that results be presented ascontinuous time series, as well as seasonal averages.The addition of climatology to Fire Family was sug-gested. Participants also highlighted the need for betteraccess to non-climatic data, such as fire starts, histori-cal fire data, and escaped/prescribed burn data. Theyrecommended that these data be synthesized for fire-climate research.

Overall Remarks. A key focus of the participants’ re-marks was that fire management and the relationshipbetween fire and climate is mitigated by other impor-tant land/ecosystem management concerns, such asrestoration, range management, wildlife management,resource management and resource availability. Partici-pants acknowledged the complexity of the interrela-tionships between all of these factors and suggested

62

that database and observation network needs be put ina long-term context: What will our needs be 5, 10, 15years from now? Finally, participants suggested thatthere were far too many individual databases, spreadamong many agencies; they stressed that easy access todata was essential and they recommended interagencysharing and coordination of databases.

Information Transfer and Decision Analysis ToolsThe discussion of information transfer and decisionsupport tools focused on issues of access to informa-tion, usability of information and products (tools), re-search needs, and remarks regarding practical opera-tional concerns.

Information Access. Key issues regarding informationaccess, continued the thread of remarks about databaseneeds. Participants pointed out the need for better ac-cess to databases, including access to information atglobal, national, regional and local scales and the occa-sional context-sensitive need for access to informationoutside the fire manager’s particular region. As infor-mation to support resource allocation in both theshort-and long-term was a prime concern of workshopparticipants, access to both forecasts and historical andinformation was recommended. Participants recom-mended access to information on a variety oftimescales ranging from short-term (2-3 months), toannual, to long-term (3-10 years), to “really long-term”(pentad-decade-century). Participants pointed out thatinformation needs vary, depending upon the timescaleof decision-making. For example, short-term timescaledata fulfilled needs for operational use, crew training,and water management. Regionally specific seasonalforecasts were identified as a much needed product. Atannual and longer timescales, more qualitative infor-mation is needed for prescribed fire planning, land-scape preparation and NEPA (National EnvironmentalPolicy Act) process. Participants suggested that vari-ables such as precipitation, temperature, and indexdata were required at the annual timescale at spatialresolutions from national-to-landscape (30 m).

Usability. In order to make the aforementioned dataand information, as well as any decision support tools,most useful to fire managers participants recom-mended one-stop shopping, as much as is possible. Par-ticipants emphasized that information needs varied byissue, by job, and by time; therefore, they recom-mended that information be available at a variety ofscales and levels of analytical complexity. They stressedthat products need to have telescoping or layering fea-

tures, in order to give users the ability to move up anddown temporal and spatial scales with ease, and to giveusers the ability to specify levels of complexity. In addi-tion, participants pointed out that graphical and mappresentation is far more useful than data tables andlists. Participants recommended that decision supporttools and information transfer products that allow forinteraction and dynamic decision-making. Participantsemphasized the need to provide historical informationin order to support resource allocation, restorationtiming, and healthy ecosystems long-term planning. His-torical information was also identified as an importantmeans of justifying funding for land/ecosystem man-agement objectives in out years. With regard to the lat-ter, participants identified geographic relationshipsbased on the response to ENSO (e.g., regional bipolerelationships), and long-term trends and oscillations(e.g., the Pacific Decadal Oscillation) as phenomena ofinterest for further research.

Finally, participants in the breakout groups under-scored the need to publicize to the fire managementcommunity that present-day climate forecasting is ap-proximately as well-developed as weather forecastingwas in the 1960s. Participants agreed that there iscause for optimism with regard to the future of long-term climate and associated fire forecasting.

TrainingTraining and knowledge transfer were identified as im-portant goals by the workshop participants. Their re-marks focused on what training should consist of andhow to best achieve this kind of knowledge transfer.All breakout groups recommended more training onclimatology and on issues of the relationship betweenfire and climate.

Training Needs. Workshop participants recommendedthe development of a primer on climate-fire relation-ships, written by instructional and design experts (inconsultation with climatologists and meteorologists).They emphasized the need for crisp graphics (“a pic-ture is worth a thousand words”), and the use of videoand satellite imagery and technology. They suggestedthat the primer and other climate tutorial informationshould be easy to interpret, available on video and theInternet, and that it also be presented in a range oftraining courses such as the S190-S590 training series.In order to improve access to this information, partici-pants recommended good visualization, user-friendli-ness and the ability to keep track of and update infor-mation available at all related Internet sites. Two ap-

63


proaches suggested for transfer of climate informationto fire managers were (1) hire a group of meteorology/climatology specialists to serve as translators of climate/weather information to fire managers, and (2) train cli-mate and fire to researchers to simplify the way theyexpress information and train managers to complexifytheir way of thinking about climate issues in fire man-agement. The philosophical framework for climate in-formation training for fire managers was a model oflife-long learning accompanied by a process of con-tinual updating of information to reflect the state-of-the-art. Another key aspect of the training model wasthe need for a two-way street, i.e. training of fire man-agers in the use of climate information, as well as thetraining of climatologists in the operations of fire man-agers on both short-and long-planning horizons. Par-ticipants expressed concern that training is usually thefirst thing to be cut from their budgets.

Who To Train? In addition to requesting more training,workshop participants recommended the followingpeople as the recipients of climate training for firemanagement and information about the complex rela-tionships between climate, fire and land/ecosystemmanagement:

• on-the-ground managers• regional managers• national policy and budget managers• Congress

Participants identified new hires, fire behavior analystsand incident meteorologists as key recipients of thistraining. They recommended intensive work with per-sonnel at the Geographic Area Coordination Centers(GACCs) and National Interagency CoordinationCenter.

Finally, participants recommended a nested design fordecision support tools and climate information, suchthat all spatial and temporal scales are represented.They highlighted the need for tools and informationwith emphases on operational contexts (e.g., districtscale) and planning contexts (e.g., regional-scale forproactive/strategic planning).

Climate and Fire 2001 in the

Southwest (March)

Data and Information Transfer NeedsWorkshop participants agreed that accurate climateforecasts, especially for the period six months-one yearin advance, are needed. They highlighted the fact thatif accurate information is available by January, thensufficient time is available for additional severity fund-ing to be requested. Participants identified easy accessto historical analog data as a critical need for decision-making. Many participants noted that fire manage-ment protocol required management decisions to bemade based on data from the previous 20 years. Theysuggested, however, that data from an analogous 20years would be more appropriate for their decision-making needs. Breakout group participants suggestedthat one way to present historical data would be toshow them as a time series of recent conditions coordi-nated with a time series ensemble of past analogousyears. Thus, managers could trace recent conditionsand then see a range of possible future conditions. Ex-panding on this idea, participants suggested that re-view of past climatic conditions would enable theidentification of trigger points during the fire seasonthat would serve as prompts for fire managers to makekey decisions. They suggested further research to con-nect climate and fire danger indices with forecasts.Historical and recent data of interest to workshop par-ticipants included the following parameters: tempera-ture, precipitation, relative humidity, buildup index,energy release component (ERC), FSI, as well asdrought indices, vegetation health and wind speed andwind direction. They stressed that the language bywhich data and results are communicated needs to beunderstandable to users and communicated consis-tently; they pointed out that this is particularly impor-tant for communication of risk factors and indices.Moreover, they pointed out the need for more data onsmoke, in particular interaction of smoke with terrainand smoke transport by wind.

Participants at the March workshop also highlightedthe need for access to data on social factors. They sug-gested that intra-agency and interagency issues need tobe factored in, as well as political considerations, suchas issues regarding elected officials at local, judicial andexecutive levels.

ResearchParticipants recommended a variety of research pro-grams, including physical science and sociocultural re-

64

search with regard to fire management. The breakoutgroups recommended an overall framework for re-search, as follows:

• take advantage of the fact that at the moment theWestern Governors’ Association is extremely inter-ested in the issue of fire

• build a consortium of university researchersthroughout the West, with the goal of collabora-tion between institutions (in contrast to the usualcompetitive grant funding process)

• focus each institution’s research on their particularstrengths

• address the highest priorities in fire managementand fire-climate research within the geographic re-gion (e.g., improvements in forecasting sixmonths-one year in advance, in order to improveresource allocation decision-making)

Participants recommended research to identify triggerpoints where climate conditions push fire danger aboveor below a certain threshold. With regard to triggerpoints participants suggested that it would be useful tocompare climate and fire forecasts with burn histories.Participants recommended research that investigatesthe relationship between indices and forecasted valuessuch as burn index and ERC. In addition, they recom-mended research on the impact of climate on factorssuch as plant and fuel flammability, curing, drying,etc. Participants also recommended research on smoketransport, sinks and dispersion that takes into accountinteractions between terrain and climate conditions.

Social science research recommendations includedanalyses of regional and community sociocultural dif-ferences, in order to better understand perceptionswith regard to tolerance of smoke, political power rela-tions (in terms of preparedness and receptivity to foresthealth and restoration programs), and analyses of hu-man-caused risk factors.

With regard to deterministic models, participants rec-ommended stochastic modeling and that the level ofconfidence in model output be made explicit.

Operations, Management and Decision MakingArizona and New Mexico fire managers emphasizedthe need for sustainable multi-year budgets. Theynoted that the Western Governors’ Association has a

10-year planning horizon, and they pointed out the20-year retrospective period to suggested by the Na-tional Fire Management Analysis System. Nevertheless,they noted, budgets are only allocated one year at atime; consequently, unused funds cannot be accrued inorder to achieve longer-term objectives mandated bymulti-year planning horizons. As a corollary to this,participants pointed out that the larger and more com-plex an area is, the more lead time required to carryout programs and change existing procedures andrules. Participants recommended that climate informa-tion needs to be better incorporated into decision-making, and that it should be mandatory in prescribedburning plans, contingency plans and preparednessplans. They recommended the use of the NOAADrought Monitor, as well as current and historicalPalmer Drought Severity Index data (preferablydownscaled to 1 km2 resolution). The fire managersagreed that they were able to observe ecosystem/vegeta-tion changes on the order of around two years if therewere persistent climate conditions, such as drought.They suggested that this has important implicationsfor trying to do restoration work, because if fuels in-crease during a favorable climate years, then there aresignificant changes at ecotones; if the favorable yearsare followed by dry years, then when fuels die at lowelevation, high elevation snowpack is no longer a sig-nificant factor for fire management, as the low eleva-tion areas are still exceedingly prone to fire (cf., CerroGrande).

Training and EducationWorkshop participants heartily endorsed further train-ing in the use of climate information for fire manage-ment. They recommended training classes such as theS490 training class. They highlighted the need toknow how to incorporate new and more climate/weather information into decision-making processes.Participants also recommended that local and regionalNational Weather Service meteorologists be encour-aged to attend fire-climate workshops, in order to fa-cilitate the integration of short-term weather and long-term climate information in fire management and fireweather forecasting. They recommended that meteo-rologists be trained to serve as climate specialists forfire weather decision-making.

Of equal, and perhaps greater, importance, workshopparticipants recommended education and training forthe public. A prime concern was that the publicneeded better education with regard to existing forestconditions, and the relationship between climate, for-

65


est management, and fire. They stressed that a bettereducated public would provide increased support forpolitical maneuvering to secure adequate funding forforest management efforts. Participants noted that thepublic needs to understand that just because we have asingle year of adequate precipitation does not mean thatforest health has been restored, nor does it mean thatdrought conditions have ended. A well-informed public,participants pointed out, is particularly important formanagement at the urban-wildland interface. Partici-pants also emphasized that agency officials andCongresspeople in Washington D.C. need to hearabout the synergy between long-term climate condi-tions and the ability of the fire and ecosystem manag-ers to achieve land management objectives. They notedthat they are under a lot of pressure to get out the burnand that the expectations of agency officials andCongresspeople in Washington D.C. are unrealistic. Inorder to remedy this situation, they recommended thatcongressional aides be invited to attend future fire-cli-mate meetings.

Fire Management in MexicoA situation unique to the March workshop was the at-tendance of Everardo Sanchez Camero, a Mexican firemanagement official, at the workshop. Sr. SanchezCamero provided the following insights into fire man-agement and the need for climate information forMexican fire management. He noted that currentlyMexico contracts to agencies and Canada for climatedata. He recommended that 30-, 60-, and 90-day cli-mate forecasts would be extremely useful to Mexicanfire managers. He pointed out a change in Mexicanfire management practice toward fire suppression, incontrast with past policy, which allowed fires to burn.He suggested that technical training is necessary formanagers and firefighters working on the ground. Healso suggested that, similar to Southwest fire managers’experience with U.S. agency and political officials,state and federal level Mexican fire managers and poli-ticians needed to be informed about climate-fire rela-tionships, as well as the need for greater levels of fund-ing for training, prevention, suppression and research.In addition, Sr. Sanchez Camero suggested that peoplein different levels of management to have different per-ceptions of forest management problems; governmentofficials, he noted, tend not to realize the importanceof proactive management policy until a crisis is athand.

66

Appendix

Participant List

John AndesBLM Yuma2555 E. Gila Ridge Rd.Yuma, AZ [email protected](Fax) 520-317-3350

Gail AschenbrennerCoronado National Forest300 W. CongressTucson, AZ [email protected]

William AstorBIA - Navajo Regional OfficeBranch of Forestry, Fire MgmtP.O. Box 1060, Slot 410Gallup, NM [email protected](Fax) 520-729-5029

Diane AustinBARAAnthropology Building Room 316The University of ArizonaTucson, AZ [email protected](Fax) 520-621-9608

Russ BabiakSaguaro National Park3693 S. Old Spanish Tr.Tucson, AZ [email protected]

Willie Begay, Jr.US Forest ServiceFederal Building333 Broadway Blvd. SEAlbuquerque, NM 87102

[email protected]

Leon BenBIA Branch of ForestryPO Box 10Phoenix, AZ [email protected]

Dave BehrensArizona State Land Department2901 W. Pinnacle Peak RoadPhoenix, AZ [email protected](Fax) 602-255-4061

Jim BrennerFlorida Division of Forestry3125 Conner Blvd., Suite A, Room 160Tallahassee, FL [email protected]

David BrightNational Weather Service Tucson520 N. Park Ave, Suite 304Tucson, AZ [email protected](Fax) 520-670-5167

Tim BrownDesert Research Institute (Atmospheric Sci. Division)2215 Raggio ParkwayReno, NV [email protected](Fax) 775-674-7016

Dan CayanScripps Institution of OceanographyClimate Research Division9500 Gilman Drive, Dept. 0224La Jolla, CA 92093-0224

67


[email protected]

Gary ChristophersonDept. of Geography and Regional DevelopmentCenter for Applied Spatial AnalysisThe University of ArizonaTucson, AZ [email protected]

Bob ClarkJoint Fire Science Program3833 South Development Ave.Boise, ID [email protected](Fax) 208-387-5960

L. Dean ClarkBandelier National MonumentP.O. Box 251Los Alamos, NM [email protected] ext. 120(Fax) 520-824-3421

Stan ColoffUSGS Office Deputy Chief Biologist (MS-301)12201 Sunrise Valley Dr.Reston, VA [email protected](Fax) 703-648-4238

Andrew ComrieDepartment of Geography and Regional DevelopmentThe University of ArizonaHarvill BuildingTucson, AZ [email protected](Fax) 520-621-2889

Sue ConardUSDA Forest Service - VMPR (1 Central), R&D201 14th Street, SW (Yates Bldg. 1C)Washington DC [email protected](Fax) (202) 205-2497

Noni DaltonUSFS - Eastern Great Basin Coordination Center5500 W. Amelia Earhart #270Salt Lake City, UT [email protected]

Carrie DennettChiricahua National Monument13063 E. Bonita Canyon Rd.Willcox, AZ [email protected] ext. 120(Fax) 520-824-3421

Carl EdminsterUSDA-FS Rocky Mountain Research Station2500 S. Pine Knoll Dr.Flagstaff, AZ [email protected](Fax) 520-556-2130

Jay EllingtonSouthwest Coordination CenterFire Intelligence Section333 Broadway SEAlbuquerque, NM [email protected]

Don FalkLaboratory of Tree Ring ResearchThe University of ArizonaTucson, AZ [email protected](Fax) 520-621-8229

Calvin FarrisLaboratory of Tree-Ring ResearchThe University of ArizonaTucson, AZ [email protected](Fax) 520-621-8229

Nick GarciaBureau of Indian AffairsP.O. Box 189Mescalero, NM 88340

68

505-671-4419(Fax) 505-671-4899

Gregg GarfinInstitute for the Study of Planet Earth/CLIMASUniversity of Arizona715 N. Park Ave., 2nd FloorTucson, AZ [email protected](Fax) 520-792-8795

Sherry GerlakBARAAnthropology Building Room 316The University of ArizonaTucson, AZ [email protected] or 621- 4523

Herman GucinskiSouthern Research StationP. O. Box 2680Asheville, NC [email protected](Fax) 828-257-4313

Pete GuilbertCalifornia Department of ForestryP.O. Box 944246Sacramento, CA [email protected]

Holly HartmannDepartment of Hydrology & Water ResourcesThe University of ArizonaTucson, AZ [email protected](Fax) 520-621-1422

Jan HendrickNational Interagency Coordination Center3833 S Development AveBoise, ID [email protected]

Mike HilbrunerUSFS - Fire & Aviation Management (2NW Yates)

Sidney R. Yates Federal Building 201 14th Street,SW at Independence Ave., SWWashington, DC [email protected]

Edward HollensheadPrescott National Forest344 S. Cortex St.Prescott, AZ [email protected]

Pamela HoltUniversity of ArizonaDepartment of Geography and Regional DevelopmentTucson, AZ [email protected]

Derek HoneymanUniversity of ArizonaISPE715 N. Park Ave., 2nd FloorTucson, AZ [email protected]

Steve HostetlerUSGS, Dept. of GeosciencesRoom 104 Wilkinson HallOregon State UniversityCorvallis, OR [email protected](Fax) 541-737-1200

Malcolm HughesLaboratory of Tree-Ring ResearchThe University of ArizonaTucson, AZ [email protected](Fax) 520-621-8229

Roy JohnsonBLMOffice of Fire and Aviation3833 S. Development AvenueBoise, ID [email protected]

69


Mark KaibLaboratory of Tree-Ring ResearchUniversity of ArizonaBuilding 58, Room 204Tucson, AZ [email protected]

Victoria KhalidiThe Nature Conservancy1510 E. Ft. Lowell Rd.Tucson, AZ [email protected] ext. 3484(Fax) 520-620-1799

David KirganBureau of Indain AffairsBIA Mescalero AgencyP.O. Box 189Mescalero, NM [email protected](Fax) 505-671-4899

Douglas Le ComteClimate Prediction CenterW/NP53 World Weather Building5200 Auth Rd, Room 800Camp Springs, MD [email protected] ext. 7567

George LeechBureau of Indian AffairsBIA Fire ManagementP.O. Box 560Whiteriver, AZ [email protected](Fax) 520-338-5348

Charlie LilesNational Weather Service Albuquerque2341 Clark Carr Loop SEAlbuquerque, NM [email protected]

Mary LynchAlaska Fire ServiceFairbanks, AK 99703

[email protected](Fax) 907-356-5049

Christopher MaierNational Weather Service Office2242 West North TempleSalt Lake City, UT [email protected](Fax) 801-524-4030

Ellis MargolisLaboratory of Tree Ring ResearchThe University of ArizonaTucson, AZ [email protected](Fax) 520-621-8229

Dean McAlisterUSFS300 W. CongressTucson AZ [email protected]

Tom McClellandUSDA Forest Service - Weather Program201 14th St., SWWashington DC [email protected]

Alex McCordAZ Division of Emergency Management5636 E. McDowell Rd.Phoenix, AZ [email protected](Fax) 602-231-6226

Chuck McHughAZ Division of Emergency ManagementDept. of Emergency & Military Affairs5636 E. McDowell Rd.Phoenix, AZ [email protected](Fax) 602-231-6226

70

Don McKenzieCollege of Forest Resources,Box 352100University of WashingtonSeattle, WA [email protected]

Ron MelcherArizona State Land DepartmentFire Management Division2901 W. Pinnacle Peak Rd.Phoenix, AZ [email protected](Fax) 602-255-4061

Carmen Melendez-AllenCoronado National Forest300 W. Congress, 6ATucson, AZ [email protected](Fax) 520-670-4567

Jay MillerUniversity of ArizonaDepartment of Geography and Regional DevelopmentP.O. Box 210076Tucson, AZ [email protected](Fax) 520-621-2889

Steven MirandaUSFSP.O. Box 68Penasco, NM [email protected](Fax) 505-758-6236

Ron MoodyUSDA Forest Service/Southwest Region333 Broadway Blvd. SEAlbuquerque, NM [email protected]

Ted MoorePike, San Isabel National ForestsUSDA Forest Service

1920 Valley DrivePueblo CO [email protected]

Barbara MorehouseInstitute for the Study of Planet Earth/CLIMASUniversity of Arizona715 N. Park Ave., 2nd FloorTucson, AZ [email protected](Fax) 520-792-8795

John MortonUS Fish & Wildlife Service/Fire ManagementP.O. Box 1306Albuquerque, NM [email protected](Fax) 505-248-6475

Ron NeilsonForestry Sciences Laboratory3200 SW Jefferson WayCorvallis, OR [email protected](Fax) 541-750-7329

Rick OchoaNational Interagency Fire Center3833 S. Development Ave.Boise, ID [email protected]

Richard OkulskiNational Weather Service Tucson520 N. Park Ave, Suite 304Tucson, AZ [email protected](Fax) 520-670-5167

Barron OrrUniversity of Arizona Office of Arid Lands1955 E. 6th St. 205cTucson, AZ [email protected]

71


Jonathan OverpeckInstitute for the Study of Planet EarthUniversity of Arizona715 N. Park Ave., 2nd FloorTucson, AZ [email protected](Fax) 520-792-8795

Tom PaganoDepartment of Hydrology & Water ResourcesThe University of ArizonaTucson, AZ [email protected](Fax) 520-621-1422

Leland PellmanBureau of Indian AffairsP.O. Box 189Mescalero, NM [email protected](Fax) 505-671-4899

Dan PitterleSan Carlos Apache TribeP.O. Box 0San Carlos, AZ [email protected]

Roger PulwartyNOAA/Office of Global Programs1100 Wayne Avenue, Suite 1225Silver Spring, MD [email protected] ext. 103

Kelly RedmondDesert Research InstituteWestern Regional Climate Center2215 Raggio ParkwayReno, NV [email protected](Fax) 775-674-7016

Merrick RichmondLaboratory of Tree-Ring ResearchUniversity of ArizonaTucson, AZ 85721


Kirk RowdabaughArizona State Lands2901 E. Pinnacle Peak Rd.Phoenix, AZ [email protected](Fax) 602-255-1781

Danny SaizaBureau of Indian AffairsP.O. Box 189Mescalero, NM 88340505-671-4419(Fax) 505-671-4899

Everado Sanchez CameroSEMARNATAv. San Jose #7 e/ San Jorge y San MarcosFracc. San AngelHermosillo, [email protected] 32711

Erick Sanchez FloresUniversity of ArizonaDepartment of Geography and Regional DevelopmentTucson, AZ [email protected]

Jim SchroederGrand Canyon National ParkP.O. Box 879Grand Canyon, AZ [email protected]

John SchulteUSDA-Forest ServiceSouthwest Coordination Center333 Broadway Blvd. SEAlbuquerque, NM [email protected](Fax) 505-842-3801

72

Karl SieglaffArizona State Land Department233 N. Main Ave.Tucson, AZ [email protected](Fax) 520-628-5847

Jacqueline SmithUniversity of ArizonaDepartment of Geography and Regional DevelopmentTucson, AZ [email protected]

John SnookInteragency Fire Forecast and Warning Unit6101 Airport RoadRedding, CA [email protected]

Ken StewartTucson Fire Department4902 E. PimaTucson AZ [email protected]

Lori StilesUniv. of Arizona News ServicesThe University of ArizonaTucson AZ [email protected]

Tom SwetnamLaboratory of Tree Ring ResearchUniversity of ArizonaTucson, AZ [email protected](Fax) 520-621-8229

Lester TisinoUSFS Douglas R.D.3081 Leslie Canyon Rd.Douglas, AZ [email protected](Fax) 520-670-4588

Mitch TobinArizona Daily Star4850 S. Park Ave.Tucson AZ [email protected]

Rocky TowUSFS Santa Catalina R.D.5700 N. Sabino Canyon Rd.Tucson, AZ [email protected]

Sheri TuneCoronado National Forest300 W. CongressTucson, AZ [email protected](Fax) 520-670-4511

Miguel VillarealUniversity of Arizona25 W. 19th StreetTucson, AZ [email protected]

Bill WaterburyUSDA Forest Service/Southwest Region333 Broadway Blvd. SEAlbuquerque, NM [email protected]

Paul WerthNW Interagency Coordination Center5420 NE Marine DrivePortland, OR [email protected](Fax) 503-808-2750

Anthony WesterlingScripps Institution of OceanographyClimate Research Division9500 Gilman Drive, Dept. 0224La Jolla, CA [email protected](Fax) 858-534-8561

73


Jeff WhitneyUS Fish & Wildlife ServiceNew Mexico Ecological Services Field Office2105 Osuna NEAlbuquerque, NM [email protected] ext. 105

William A. WilcoxCoronado National Forest5990 South Highway 92Hereford, AZ [email protected](Fax) 520-670-4640

Butch WilsonBuenos Aires NWRP.O. Box 109Sasabe, AZ [email protected] ext. 104(Fax) 520-823-4247

Barbara WolfBARAAnthropology Building Room 316The University of ArizonaTucson, AZ [email protected](Fax) 520-621-9608

Klaus WolterNOAA-CIRES / Climate Diagnostic Center325 BroadwayBoulder, CO [email protected]

Steve YoolDept. of Geography and Regional DevelopmentThe University of ArizonaTucson, AZ [email protected](Fax) 520-621-2889

Gary ZellNational Weather Service Tucson520 N. Park Ave, Suite 304Tucson, AZ 85719


74

Agenda: Fire & Climate 2001

February 14-16, 2001Four Points by Sheraton Tucson University Plaza1900 E. Speedway BoulevardTucson Arizona

Wednesday February 14

Morning Weather, Climate, Fire The 2000 Fire Season and Beyond8:15-8:45 Welcome and Introductions – Barbara Morehouse and Tom Swetnam8:45-9:15 Climate and Fire: Framing the Issues in the Context of the 2000 Fire Season – Tim Brown9:15-9:45 Weather and the 2000 Fire Season – Rick Ochoa10:00-10:30 NIFC Resource Impacts During the 2000 Fire Season – Rick Ochoa10:30-11:00 Climate Forecasts for 2001 – Klaus Wolter11:00-11:30 Fire Forecasts for 2001 – Tim Brown

Afternoon Weather, Climate, Fire The 2000 Fire Season and Beyond (continued)1:15-1:45 Climate Information: Recent Developments in Data Access at WRCC – Kelly Redmond1:45-2:15 The Cerro Grande Fire - John Snook

Beyond the 2000 Fire SeasonMAPSS: Mapped Atmosphere Plant Soil System – Ron NeilsonAn Overview of the Joint Fire Science Program and RFPs for 2001 – Bob Clark

2:15-3:15 Open Discussion: Prescribed burns after Cerro Grande3:30-5:00 Open Discussion: Lessons from the 2000 fire season: What can we do better in 2001?

Thursday February 15

Morning Climate Prediction8:00-8:20 Climate Modeling Overview – Klaus Wolter8:20-8:40 Climate Modeling and Diagnostics – Dan Cayan8:40-9:00 An Experimental Seasonal Fire Weather Forecast for the Western U.S. – Anthony Westerling9:00-9:45 Climate Prediction: ENSO versus non-ENSO Conditions – Douglas LeComte10:00-12:15 Climate Prediction Interpretation and Evaluation Workshop – Holly Hartmann and Tom Pagano

Afternoon The Use of Climate Information by Fire Managers1:30-2:15 Open Discussion: Identification of Issues around the use of Climate Information by Fire Managers2:15-5:00 Breakout Groups

Friday February 16

Morning Integrated Assessments8:00-8:30 Interagency Integrated Assessments – Roger Pulwarty8:30-9:00 Tools to Improve Fire Risk Management – Barbara Morehouse and Steve Yool9:00-9:30 An Overview of CEFA and its Tools for Fire Managers – Tim Brown9:30-10:00 Climate, Fire and the Need for a National Climate Service – Jonathan Overpeck10:00-12:00 Discussion

75


Agenda: Fire & Climate in the Southwest 2001

March 28, 2001Arizona Ballroom, Student Union BuildingThe University of ArizonaTucson, Arizona

8:30-9:00 Introductions and Overview

9:00-9:40 Climate of the Southwest — Andrew Comrie, Department of Geography and RegionalDevelopment, University of Arizona

9:40-10:10 Fire, Climate and the 2000 Fire Season in the Southwest — Tom Swetnam, Laboratory ofTree-Ring Research, University of Arizona

10:10-10:40 Fire and Emergency Management in Arizona – Alex McCord and Ron Melcher, ArizonaDepartment of Emergency Management, and Arizona State Land Department

10:50-11:20 CEFA Climate Products for Fire Managers — Tim Brown, Program for Climate, Ecosystem andFire Applications

11:20-11:50 EPA Decision Support System for Fire Managers — Barbara Morehouse and Steve Yool,CLIMAS and Department of Geography and Regional Development, University of Arizona

11:50-12:10 Discussion

1:10-1:40 Fire History in the Jemez Mountains: The Bandelier Perspective – L. Dean Clark, National ParkService, Bandelier National Monument

1:40-2:30 2001 Climate Forecasts for the Southwest — Tom Pagano, Department of Hydrology and WaterResources, University of Arizona

2:30-2:50 Discussion of Fire Management Climate Information Needs

3:00-5:00 Breakout Groups

climas...climas • institute for the study of planet earth climate assessment for the southwest...

Documents