Developing an improved wildland fire i i i temissions inventory

Sim Larkin, Tara Strand, Robert Solomon, USFS AirFire Team, PNW Research StationStacy Drury Sean Raffuse Neil Wheeler Ken Craig Lyle Chinkin Sonoma Tech IncStacy Drury, Sean Raffuse, Neil Wheeler, Ken Craig, Lyle Chinkin, Sonoma Tech, Inc.Nancy Martinez, Harvard; Pete Lahm, USFS WO

EPA Emissions Inventory ConferenceSeptember 29 2010September 29, 2010San Antonio, TX

Towards an improved EITowards an improved EI

Increasing Demand for Fire EIIncreasing Demand for Fire EIMany groups, not just EPA:

–land agencies, states, first nationsFaster EI development:Faster EI development:

–yearly reportingregional/national

l t t lMultiple purposes:–GHG/Carbon reporting

annual total

p g–Short lived climate forcers

Air quality–Air quality localized

daily/hourly

Current Fire EIsCurrent Fire EIs

• Many Fire EIs done for many purposesMany Fire EIs done for many purposes• Methodologies differ

Diff t Fi D t t– Different Fire Detects– Different Fuel Loadings

ff C– Different Consumption Models– Different Emissions Factors– databases and models change over time

• Resulting differences can be large– reasons not generally clear

Current Fire EIsCurrent Fire EIs

• Resulting differences can be largeResulting differences can be large– reasons not generally clear

250

300

ions

(Tg/

yr) Wiedenmyer and Neff

2008 WF NEIU.S. EPA UNFCCCRanderson et al.

150

200

re C

O2

Em

iss

50

100

150

48 W

ildla

nd F

i

0

50

2003 2004 2005 2006

Low

er 4

Cross-cutting model intercomparison projectintercomparison project

Evaluations at many different output levels

Anyone can add data/model output

Standard test cases

Standard comparison metrics

All documented on web: http://semip.org

Test CasesTest CasesSEMIP utilizes a series of test cases to analyze to serve as a starting point for SEMIP'sanalysis These case have been selected to try to represent a wide array of conditions fire types

National Cases1 Fires Everywhere 2 National

analysis. These case have been selected to try to represent a wide array of conditions, fire types,regions, vegetation, weather, and use cases. It is expected that additional test cases will be added overtime.

1. Fires Everywhere 2. National Emissions Inventory 2008

Regional Wildfire Case3 C lif i Wildfi (200 & 2008)3. California Wildfires (2007 & 2008)

Individual Wildfire Cases4. Bugaboo Complex 2007 5. Tripod4. Bugaboo Complex 2007 5. Tripod

Complex 2006

Prescribed Fire Cases6 N th t R i l R S 76. Northwest Regional Rx Season 7.

Southeast Regional Rx Season 8. Naches, WA Rx Fire

Emissions Bulk FormulaEmissions Bulk FormulaAREA FUEL

LOADBURNING

EFFICIENCYEMISSIONS

FACTORx x xEMISSIONS = AREA LOAD EFFICIENCY FACTORx x x

CONSUMPTION

?1 ? ? ?certa

inty

? ? ? ?

Unc

A priori, might expect uncertainties to rank as:Area < Consumption < EF

Preliminary resultsPreliminary resultsfrom literature searchCourtesy N. Martinez

Not

S

peci

fied

Preliminary resultsfrom literature searchCourtesy N. Martinez

dN

ot

Spe

cifie

d

Fuels and Consumption

Example: Tripod Fire 2006

175,000+ acres(71,000 hectares)( , )

Multiple fuel typesMultiple fuel types

ICS Report location in f b dcorner of area burned

How does point variability p yscale to a large fire area?

old new

old new

Scaling from a point to westwideScaling from a point to westwide

e

☐

mpt

ion/

Acr

eed

ian)

☐

al C

onsu

miz

ed to

me

☐

rage

d To

ta(n

orm

ali

Aver

od oc od ve Ave

OR

/Av

e

ern

Ave

Trip

oR

epor

t L

Trip

oA

rea

Av

WA

A

WA

/OID

/MT

A

Wes

te A

Based on 12 model pathwaysTripod Average also includes high-res fuels -> CONSUME (squares)

Changes in 2008 Due to PathChanges in 2008 Due to Path• Only large fires (from MTBS)• Fuel maps: FCCS1-1km FCCS2-30m LANDFIRE-30m fuels• Fuel maps: FCCS1-1km, FCCS2-30m, LANDFIRE-30m fuels• Consumption & Emissions: CONSUME3.0 & FOFEM5.7

Fire InformationFire Information • Emissions inventory y

depends explicitly on fire area, location, and timing

• Major regional differences in reporting systems, firein reporting systems, fire size, fire types, and fire detection

Quantity and TimingChoice of fire information system affects timing

Quantity and Timing

Bl ICS209 Wildfi R tBlue = ICS209 Wildfire ReportsRed = MODIS Fire DetectsYellow = HMS Fire Detects

2003-2006 Average Acres Burned

1.6 ICS209

0.8

1.0

1.2

1.4

on a

cres

)

MO DIS

SMARTFIRE

0.0

0.2

0.4

0.6

Are

a (m

illi

0.0

Janu

ary

Februa

ry

March

April

May

June Ju

ly

Augus

t

Septem

ber

Octobe

r

Novem

ber

Decem

ber

Fire InformationFire Information

Green = EPA NEI fires missing from gNC DFR database

Red = NC DFR fires missing from

Courtesy G. Curcio

EPA database

The West• WRAP states 2002 inventory compared with various firecompared with various fire reporting systems (both ground and satellite-based).

• Most systems only reported a fraction of the WRAP EI acres.

• SMARTFIRE overall detected• SMARTFIRE overall detected only 52% of WRAP EI acres.

• But, SMARTFIRE does not include Ag burns (~1/3 of total).

Soja et al, 2009

Some Preliminary ConclusionsSome Preliminary ConclusionsRelative weight of uncertainties:

Area ≈ Consumption ≈ best known EFs

f i l ifor regional summaries.

But, for air quality uses, the uncertainty in area becomes even more important.

Caveat: in much of this, we don’t know what we don’t know. This is especially true for EFs.

Plume Rise

We matched 163 plumes.

• MISR dynamic range less• MISR dynamic range less than modeled

• Model underpredicting small fires, overpredicting l filarge fires

• Clear regime difference in MISR data between West and Southeast

21

• Poor performance throughout the West

Changes UnderwayChanges UnderwayFocus on:

• Plume height constraints• Additional emission calculation schemes• Better treatment of fuel moisture• Spatially explicit fuels (all fuels in perimeter)• More recent fuel map (FCCS2 30m national)• Incorporating more local data

Fire information e g from states• Fire information, e.g. from states• Local fuel surveys

NeededNeededMore local data:

f l•fuels•fire perimeters•management information•ancillary data

More types of information:More types of information:•cross comparison of various types of data may be able to constrain uncertainties

More explicit treatment of uncertainty

Aggregation databases (gathering information in one spot)

BlueSky FrameworkFireInfo

BlueSky FrameworkRed = Current NEI path

Fuels

Total

SMARTFIREICS-209Rx SysManualOther FCCS Total

Consumption

TimeRate

OtherNFDRSHardyLANDFIRE*Ag*Other

CONSUME 3FOFEMFEPS Rate

Emissions

Pl

FEPSEPMClearSky (Ag)*Satellite*Other

Rx / WFFEPSEPMWRAP FEPS Plume

Rise

Dispersion /Trajectories

IdealizedManualOther

FEPSLiterature*EPMFOFEMOther Briggs

Multi-core**D k ** TrajectoriesDaysmoke**Other CalPuff

HYSPLITCMAQ

GOES-10 and -12:

GOES NOAA-15,-16, and -17: POES

Terra and Aqua: MODIS

Global Fire Detects

Ground ReportsGround Reports

A H

azar

d ng

Sys

tem Fire Pix

& Sub

ParamWF

BlueSky Improvements Underway

NEWSMARTFIRE

TRMM: TMI, PR; DMSP: SSM/I; Aqua: AMSR-E;

17: POES

Terra and Aqua: MODIS Fire Detects

GOES-11 and 12: GOES

Western Hemisphere Fire DetectsFIRE INFO

(SMARTFIRE)

stur

e2R

T)

NO

AAM

appi

n

xel Location -Pixel Fire eterization

F-AB

BA

SSM/I; Aqua: AMSR E; NOAA: AMSU-BRainfall

Suite of GEO Satellites carrying infrared sensorsRainfall

FUELS

Dea

d Fu

el M

ois

TMP

A-R

T (3

B42

data

set FU

ELS

FIRE R

FLAM

BE’ PA

T

NASA/NOAA Earth

Key

Terra/Aqua: MODISDirect Broadcast

CONSUMPTIONive

Fuel

M

oist

ure

East

fire

Algo

rithm

EMISSIO

NS

RADIATIVE PO

WER

THW

AY

FUEL MOISTURE

NASA/NOAA Earth science results

NEW COMPONENTS AND CONNECTIONS ARE IN RED

Terra: MISRStereo Heights

CONSUMPTION

EMISSIONS

L M E A

R

eigh

tin

tsas

et

g

Unknown: MISP (Future)Stereo Heights

Terra/Aqua: MODISFire Radiative Power

GOES-R: ABI (Future)Fi R di ti P

NPP: VIIRS (Future)Fire Radiative Power

DISPERSION/

Fire Radiative

Power

MIN

X dataset

PLUME RISE

Plum

e H

eC

onst

rai

MIN

IX d

at

Fire Radiative PowerDISPERSION/CHEMISTRY

e

CURRENT BLUESKY &SMARTFIRE (BSF-SF)

Tackling the Fire Information ProblemSMARTFIRE 2

Raw fire information sources

Any number of input databases

• ICS-209Source-specific association

Fires in similar conceptual dimension for each data source

ICS 209• NOAA HMS• MBTS (burn scar)• Helicopter perimetersdimension for each data source

Spatio-temporal association

p p• FETS• State Rx DBs• Annual reports

Query-able database of associated fire information

•Local forest databases, e.g. fuels

Preserves all input dataReconciliation and other task-specific processing

Preserves all input data

Specialized output streams (e.g., real-time daily activity, retrospective, geographic, agency specific)

Customizable MetricsSMARTFIRE 2

Query-able database of associated fire information

Reconciliation and other task-specific processing

Specific outputs• probability of existence• lat, lon point location

Specialized output streams (e.g., real-time daily activity, retrospective, geographic, agency specific)

lat, lon point location• spatial extent• final size• time resolved growthagency specific)• fuels• emissions• moreMultiple processing streams

Each processing stream based on a customizable set of “trust” metrics for each i t dbinput db

Thank youThank you

In particular: JFSP NASA DOI/USFSIn particular: JFSP, NASA, DOI/USFS

Also: NFP, USFS, EPA and others

More information:• SEMIP: http://semip.orgg• BlueSky & SMARTFIRE: http://blueskyframework.org

Sim Larkinlarkin@fs.fed.us206-732-7849

End of Presentation

Thank you.

Preliminary resultsPreliminary resultsfrom literature searchCourtesy N. Martinez

Not

S

peci

fied

Not

S

peci

fied

ScalingScalingExample: Tripod Fire 2006

175,000+ acres(71,000 hectares)( , )

Multiple fuel typesMultiple fuel types

ICS Report location in f b dcorner of area burned

How does point variability p yscale to a large fire area?

SMARTFIRE FireEvent Development (1 of 2)

S. Raffuse, Sonoma Tech

SMARTFIRE FireEvent Development (2 of 2)

S. Raffuse, Sonoma Tech