(l-)2.. source attribution during whitex a modeling study ·  · 2009-09-23source attribution...

14
.. SOURCE ATTRIBUTION DURING WHITEX A MODELING STUDY Roger A. Stocker and Roger A. Pielke Atmospheric ScienceDepartment Colorado State University Fort Collins, CO 80523 Abstract The application of deterministic mesoscale models to identify sources of pollution provides an effective new tool in regulatory management. Such models provide spatially and temporally varying fields of wind and turbu- lence as forced by local terrain, surface heating patterns, and background flow. Lagrangian particle dispersion model integrations can then be used to estimate the resultant distribution of pollution from point and area sources. The deterministic modeling work presented in this paper, representing the February 11-12, 1987high sulfur event during WHITEX, shows that the Navajo Generating Station plume was a major contributor to the high sulfur during this period. This work also suggeststhat there was a small contribu- tion from a northern source and a small contribution from the smelters in southern Arizona at night. Introduction and Approach During the winter of 1987, a study was conducted to determine the impact of Navajo Generating Station's (NGS) effluent plume on the local area surrounding the power plant. This six week study, the Winter Haze In- tensive Tracer Experiment (WHITEX), involved measurements of both me- teorological and particulate variables to determine if NGS emissionsplayed a major role in the genesisof the winter time haze episodesthat have been documented in the area.! There are several ways to approach the question of NGS's relative contribution to the haze episodes (i.e., statistical, deterministic modeling, etc.). The approach used in this work involves deterministic modeling. To this end, a casestudy was picked during WHITEX in which an extreme haze episodeexisted. The case study picked was one from February 9-14, 1987. Figure 1 shows the time series in Julian days (JD) of sulfur measurements for the WHITEX study for a number of different sites throughout the area. Although this episode does not represent the event with the highest sulfur reading, it is the event with the longest duration and greatest area coverage. The limitations of the particular model used in this study (i.e., no up- dating of synoptic condition and homogeneous initialization) suggest that only a portion of this haze episode can be studied in one continuous sim- ulation. From an analysis of the National Meteorological Center surface weather maps for this period, it was determined that during the February From .Visibility and Fine Particles," a Transactions of the Air & Waste Management Association edited by C. V. Mathai (1990)

Upload: buixuyen

Post on 01-Apr-2018

215 views

Category:

Documents


1 download

TRANSCRIPT

(l-)2..

SOURCE ATTRIBUTION DURING WHITEXA MODELING STUDY

Roger A. Stocker and Roger A. PielkeAtmospheric Science Department

Colorado State UniversityFort Collins, CO 80523

Abstract

The application of deterministic mesoscale models to identify sourcesof pollution provides an effective new tool in regulatory management. Suchmodels provide spatially and temporally varying fields of wind and turbu-lence as forced by local terrain, surface heating patterns, and backgroundflow. Lagrangian particle dispersion model integrations can then be used toestimate the resultant distribution of pollution from point and area sources.

The deterministic modeling work presented in this paper, representingthe February 11-12, 1987 high sulfur event during WHITEX, shows that theNavajo Generating Station plume was a major contributor to the high sulfurduring this period. This work also suggests that there was a small contribu-tion from a northern source and a small contribution from the smelters insouthern Arizona at night.

Introduction and Approach

During the winter of 1987, a study was conducted to determine theimpact of Navajo Generating Station's (NGS) effluent plume on the localarea surrounding the power plant. This six week study, the Winter Haze In-tensive Tracer Experiment (WHITEX), involved measurements of both me-teorological and particulate variables to determine if NGS emissions playeda major role in the genesis of the winter time haze episodes that have beendocumented in the area.!

There are several ways to approach the question of NGS's relativecontribution to the haze episodes (i.e., statistical, deterministic modeling,etc.). The approach used in this work involves deterministic modeling. Tothis end, a case study was picked during WHITEX in which an extreme hazeepisode existed. The case study picked was one from February 9-14, 1987.Figure 1 shows the time series in Julian days (JD) of sulfur measurementsfor the WHITEX study for a number of different sites throughout the area.Although this episode does not represent the event with the highest sulfurreading, it is the event with the longest duration and greatest area coverage.

The limitations of the particular model used in this study (i.e., no up-dating of synoptic condition and homogeneous initialization) suggest thatonly a portion of this haze episode can be studied in one continuous sim-ulation. From an analysis of the National Meteorological Center surfaceweather maps for this period, it was determined that during the February

From .Visibility and Fine Particles," a Transactions of theAir & Waste Management Association edited by C. V. Mathai (1990)

'.

11-12 (JD 42-43 in Figure 1) time period, the haze around the area and espe-cially around the Grand Canyon was mostly associated with a local source.This was due to a strengthening of the high pressure and the lack of any syn-optic southeasterly transport from the smelter region in southern Arizona.2Therefore, the February 11-12 (JD 42-43 in figure 1) time period which rep-resents the highest sulfur readings during this event, was picked to try andidentify the relative contribution of both local (NGS) and regional sources.

Modeling results

In the past few years, there have been a number of simulations in-volving evaluations of thermal mesoscale circulations in the WHITEX studyarea. Most of this work involved 2-D simulations to determine the sourceof an airmass within the Lake Powell drainage basin during selected synop-tic weather events.3,4,5 This work suggests that there is little entrainmentof an airmass associated with a distant source during extended polar highstagnation events.

A 3-D large scale (western United States domain) simulation6 of thecase study being examined in this paper suggests that synoptic scale forcingis not the only transport mechanism for the movement of a plume originatingat NGS and traveling towards the southwest. An observational study7 ex-amining the wind fields around the Lake Powell area suggest that lake/landcontrasts produce local circulations with the flow down Glen Canyon duringthe day with up-Canyon flow during the night.

The following sections describe the current work using a primitiveequation mesoscale models to describe the meteorology for the case beinganalyzed. The meteorological fields are then used as input to a lagrangianparticle model9 which is applied to visualize both the impact of local andregional sources on the Grand Canyon region. Both of these models havebeen tested previously for a wide range of mesoscale and synoptic conditionsincluding sea/lan~ breez,es, lake effect storms) mountain valley flows andthermally forced cIrCUlatIons to name a few.l0,11,12,13,14,15,16,17

The objective of the modeling work presented here is to obtain a phys-ical understanding of the complex mesoscale circulations that develop in theWHITEX region along with a qualitative understanding of the impact onthe Grand Canyon of particular source regions during the case study be-ing examined. Since the resolution of the meteorological model used in thisstudy was too coarse to resolve the Grand Canyon (an area with importantsmall-scale transport circulations), only visualizations of the relative contri-butions to Grand Canyon particulate measurements from different sourceareas are presented.

832

Model Initialization

Outlined below are some of the input parameters and their correspond-ing values used in obtaining the meteorological fields for the current case

study.1. Grid size: 69 X 69 horizontal grid points (, 380 km X 380 km)j 18

vertical grid points (9 km model top).2. Time step: 60 seconds.3. Mid-level clouds covered 70% of the domain and are used to achieve an

appropriate radiational balancel8. No precipitation mechanisms.4. A constant albedo of 0.5 was obtained from satellite imageryl9,20 for

the whole area and is assumed to be associated with snow cover on thesurface and mid-level clouds.

5. An initial temperature and moisture lapse rate of 7.3°C km-l, 3.7°C km-1 and 17.2°C km -1 ( stratosphere), and 3.0 g kg-I, 1 to 2 g kg-l and0.5 g kg-l for the layers surface to 1.5 km, 1.5 km to 7.0 km, and 7.0to 9.0 km, respectively.

6. A synoptic wind of 2.0 m s-1 from the northeast imposed on the layerfrom the surface to 700 m above the lowest model elevation (, 150 mabove Page) with the synoptic wind linearly decreasing to zero at 1.5km. This implies that there is no mixing of the synoptic flow above theplanetary boundary layer (PBL) with flow within the PBL. This is veryreasonable considering the strong stability during this time as seen fromthe soundings.

7. Model simulation duration: 30 hours. The first 6 hours are used as amodel adjustment period with the last 24-hours producing the desiredfull diurnal cycle. (February 11-12 are assumed to be identical from asynoptic point of view.)Figure 2 shows the modeled terrain which makes up the WHITEX

area. The terrain elevations are in decameters with the Colorado Riverindicated by the dashed line. Specific measurement sites are also indicatedin the Figure and are represented by the numbers 1 through 9. The terr~nwithin the domain is quite complex as can be seen by examining Figure 2.The dominant vegetation in the area is Juniper-Pinyon woodland, but theimpact of mesoscale circulations due to vegetation would be small duringthis time due to the observed cold temperatures and therefore, a lack ofsignificant stomatal transpiration. As one would imagine, the mesoscalethermally forced circulations which are produced in this area are highlycomplex. The next section describes the model fields obtained for the 24-hour simulation.

Results From the Mesoscale Meteorological Model

The 30-hour mesoscale simulation of the February 11-12, 1987 casestudy was begun at 0400 1ST in the morning. Figures 3a-c show some ofthe meteorological fields for 1500 1ST. Figure 3a shows a plan view of the

833

vector surface winds which are strongly influenced by the mesoscale circu-lations. Notice that during the afternoon, the winds along Marble Canyonand towards the southeast are primarily out of the northeast which indi-cates down valley flow. This is in agreement with surface winds obtainedin this area for WHITEX during this time period.2 This northeast flow ismostly due to the heating of the Kaibab Plateau which acts as a local heatsource during the day and drives daytime upslope winds. In contrast, the500 m ("",the effective stack height of the NGS plume) vector wind fields alsosuggest northeast transport, but these winds are primarily being driven bythe synoptic pressure gradient force imposed at the start of the simulation.Another interesting feature in the 500 m flow is the cyclonic eddy circula-tion which develops in the mountain/valley region around Monticello (seeFigure 2). This circulation would tend to block any flow coming east alongthe San Juan River basin. This feature is evident at both the surface and500 m levels. Figure 3c shows the thermal structure along a y-z cross sec-tion centered at 111.7°W. In this Figure, the region around Page and NGS(""'36.9°N) shows a PB1 of approximately 700 m. The deepest PB1 occursin the mountain range around Bryce Canyon where the PB1 exceeds 1.5 kmdue to strong mesoscale circulations in the area. Observations obtained fromthe Canyonlands and Page soundings for this time period suggest that themaximum PB1 depth around these two areas was approximately 800-900 m.The model produces maximum PB1 depths in these two areas on the orderof 1 km. The PB1 can be used as a measure of the vertical depth of mixingin the atmosphere.

Figures 4 a-c represent the same figures as Figures 3 a-c except theseare representative of the meteorology at 0300 1ST in the morning. At thesurface, the vector winds are quite different from the afternoon plots. Atthis time, the nocturnal circulations have developed fully and show the kata-batic winds off the Kaibab Plateau and the mountains around Bryce Canyon.The surface winds from the Grand Canyon towards the Marble Canyon showmostly northwest winds. At Page, the surface winds are out of the ~outh-west. This agrees well with the observations taken for this time duringWHITEX. These Page winds coupled with north winds to the north wouldsuggest stagnation around Page at night. The 500 m winds do not correlatesignificantly with the surface winds at night due to the lowering of the PB1at night and therefore, the decoupling of these two levels. These northeastwinds are still primarily being forced by the synoptic winds in the model.This suggests that at night, any NGS emissions would still travel towardsthe Grand Canyon, but not impact the surface. The thermal structure inFigure 4c illustrates that this decoupling is due to the strong radiationalInVerSIon.

Lagrangian Particle Model Results

This section will involve the results of three different model runs. Thefirst is a continuous particle release at the effective stack height of NGS

834

I

starting at 1000 1ST and continuing for 24 hours. The second and thirdruns involve a group of volume sources released along a line at 0920 1STand 1900 1ST to determine the impact of local versus regional sources. Dueto the cost involved with calculating concentration values using a Lagrangianframework and the coarse resolution of the meteorological model (discussedpreviously), absolute concentration values will not be calculated. However,a bulk estimation will be obtained to generate a qualitative estimate ofthe relative impact of each source area. This bulk estimate is obtained bycounting up the total number of particles that pass over the Grand Canyonregion from each source area in the second two 24 hour simulations. Theasterisks in the figures represent the same sites as those shown in Figure 2.

Figures 5 a-d show plan views of particles representing the NGS tracerwith the particles being released at a rate of 18 particles per hour. Thesefigures show that when one uses the synoptic conditions established in thelast section with the inclusion of thermally forced mesoscale circulations,the NGS plume impacts Hopi point in about 12 hours. The plume tends totravel on the east side of the canyon rim which agrees with the time-lapsesequences of this event. At 0200 1ST, notice that the last half of the plumeis very tightly clustered due to the lack of any appreciable turbulence atnight. Therefore, this part of the plume stays above the surface during thenight due to the decoupling and is starting to be mixed down to the surfaceby 0900 the next morning.

The next two figures represent visualizations of the impact of both re-gional and local sources when there is not a significant amount of stagnationevident along the transport path. For this situation, there would be littleaccumulation of the pollution. Therefore, the airmass into which the plumemixes along the transport path is assumed to have a much smaller pollutionconcentration than the plumes themselves. This can be somewhat justifiedfor this case due to the strong winds that develop in the region resulting invery little stagnation of the plumes in the particle simulations shown here.

Figure 6 a-d represents a 24 hour simulation intended to assess theimpact of regional as well as local sources. This simulation assumes thatthere was not a significant amount of stagnation evident along the transportpath. Therefore, the air in which the plume mixes into along the path isassumed to have a much smaller pollution concentration then the plumesthemselves. This can be somewhat justified due to the strong winds thatdevelop in the region and the fact that very little of the plumes stagnate inthe simulations. To do this, four line sources of different width but the samenumber of particles (300) are released at 0920 1ST through a depth of 800m starting at the surface. There is a northern border source, represented bya' /' sign, to determine the impact of the source(s) to the north of Canyon-lands. There is an eastern border source, represented by a 'I' sign, in the SanJuan River basin to determine the impact of the Four Corner's power plant.The southern source, indicated by the 'I' sign, represents the smelters insouthern Arizona and there is also the NGS contribution represented by a'+' sign between Page and NGS.

835

..

By 1500 1ST, the smelter contributions have moved toward the southand are almost completely removed from the area. The NGS componentis duplicating what was seen with the continuous release as expected. Thenorthern source is also moving south down the Colorado River basin. TheFour Corner's contribution is not moving much towards the east due to thecyclonic eddy circulation discussed previously. At 2100 1ST, the smeltercontribution is no longer within the domain. The northern source is verydiffuse due to the strong mesoscale circulations which dispersed the plumeas it came down the Colorado River basin. The Four Corner's componentis moving towards the south away from Page and the Grand Canyon. By0300 1ST, the northern source's leading edge has almost reached the GrandCanyon, but it is very spread out. The Four Corner's contribution has trav-eled more toward the southwest but is still not affecting the Grand Canyon.At 0900 1ST, 24 hours after the initial release, the northern source has fi-nally impacted the Grand Canyon, but it is so diffuse that any contributiondue to a northern source for this WHITEX even would be expected to becomparatively small.

Figures 7 a-d are shown to represent what the different sources inFigures 6 a-d would contribute if the plumes were released at night (19001ST). During this sequence of figures representing the nocturnal flow, themost interesting feature is the separation of the NGS plume (represented bythe particles) with the majority of the plume still traveling toward NGS, butthere is a significant amount of the particles which stay around the Page areaduring the entire period. This is due to the katabatic winds at night fromterrain features around Page which have been discussed previously. Anotherimportant result is that by 0900 1ST, the southern Arizona contribution,due to local mesoscale circulations, has mixed with the NGS contribution inthe southwest corner of the domain. This could explain some of the smeltercontribution which was evident at the Grand Canyon during this time (thesmelter contribution was about half the NGS contribution during this timeperiod according to statistical models)2 but due to presence of the Canyon,which was not resolved in this simulation, it is impossible to distinguishbetween an active contribution and one which might have stagnated withinthe Canyon.

There is observational evidence to support the distribution shown inboth Figures 6 and 7. This evidence comes from an examination of the con-tour plot of sulfur for February 11, 1987 (see figure 6.23 of the WHITEXreport2). In this Figure, there is a primary maximum around Page, Arizonaand a secondary maximum around the Green River site with a relative min-imum between the two. This type of distribution is supported by our workwith the NGS plume contributing primarily to the Grand Canyon regionand the northern source primarily impacting only the Canyonlands regionso that there would be a relative minimum concentration between Page andCanyonlands.

836

..'Concl "usions

This case study representing the high sulfur event during February11-12, 1987 of the WHITEX field study, shows that the NGS plume duringthis time was a major contributor to the sulfur concentration at the GrandCanyon. It is estimated that the travel time between NGS and the GrandCanyon was on the order of 12 hours. Also, the northern source does impactthe Grand Canyon after 24 hours, but the plume is simulated to be so spreadout for this WHITEX event that its contribution to the concentration atthe Grand Canyon would be quite small. Four Corner's impact upon theGrand Canyon is concluded to be negligible. The smelter contribution froma daytime plume is also assumed to be negligible, but during the night time,due to local mesoscale circulations, the smelter plume could combine withthe NGS plume to have some impact on the Grand Canyon.

Currently, a more complete mesoscale meteorological model is beingutilized (the Regional Atmospheric Modeling System) which will allow us toresolve the Grand Canyon in more detail as well as the larger scale synopticflow. This will allow us to produce quantitative estimates of the concentra.-tion fields to compare with observations.

Acknow ledgements

This work has been funded by the National Park Service under Grant#NA81RAHOOOOl, Amendment 17, Item 15. The computer'simulationswere done at the National Center for Atmospheric Research which is spon-sored by the National Science Foundation. Thanks are due to Dallas Mc-Donald, Tony Smith, and Bryan Critchfield for their preparation of this

manuscript.

References

1. P.S. Bhardwaja, D.W. Moon, and G.L Powell, "Visibility trends in thevicinity of an isolated power plant", Atmas. Environ., 15, 1943, 1981.

2. W. Malm, K. Gebhart, D. Latimer, T. Cahill, R. Eldred, R. Pie!ke,R. Stocker, and J. Watson, The National Park SeM1ice Report onthe Winter Haze Intensive Tracer Experiment, Draft Final Report,National Park Service, Fort Collins, CO, 1989.

3. M. Segal, C.-H. Yu, and R.A. Pielke, "Model evaluation of the impact ofthermally induced valley circulations in the Lake Powell area on long-range pollutant transport", J. Air Pollute Control Assoc., 38, 163,1988.

4. M. Segal, C.-H. Yu, R. W. Arritt, and R.A. Pielke, "On the impact ofthermally induced circulations on regional pollutant transport", Atmas.Environ., 22, 471, 1988.

5. C.-H. Yu and R.A. Pielke, "Mesoscale air quality under stagnant synopticcold season conditions in the Lake Powell area", Atmas. Environ., 20,

1751,1986.6. R.A. Stocker, R.A. Pielke, and C.J. Tremback, "A preliminary compar-

ison of the WHITEX field study with synoptic-derived trajectory re-

837

suIts", Presented at the APCA 81st annual meeting, Dallas, TX, June19-24, 1988.

7. R.C. Balling and J .1. Sutherland, "Diurnal and seasonal wind directionpatterns within the mountain-valley terrain near Glen Canyon Dam,Arizona", J. Appl. Meteor., 27, 594, 1988.

8. Y. Mahrer and R.A. Pielke, "A test of an upstream interpolation tech-nique for the advection terms in a numerical mesoscale model", Mon.Wea. Rev., 106, 818, 1978.

9. R. T. McNider, Investigation of the impact of topographic circulationson the transport and dispersion of air pollutants, Ph.D. Dissertation,University of Virginia, Charlottesville, VA, 1981.

10. S. Cautenet, and R. Rosset, "Numerical simulation of sea breezes withvertical wind shear during the dry season at Cape of Three Points, WestAfrica", Mon. Wea. Rev., 117, 329, 1989.

11. S.R. Pease, W.A. Lyons, C.S. Keen, and M. Hjelmfelt, "Mesoscale spi-ral vortex embedded within a Lake Michigan snow squall band: Highresolution satellite observations and numerous numerical simulations",Mon. Wea. Rev., 116, 1374, 1988.

12. R. W. Arritt, "The effect of water surface temperature on lake-landbreezes and thermal internal boundary layer", Bound-Layer Meteor.,40,101,1987.

13. R.C. Kessler and S.G. Douglas, Numerical simulation of mesoscaleairflow in the south central coast air basin, Draft Final Report SYSAPP-89/077, Prepared for Sigma Research Corporation, Lexington, MA,1989.

14. J .R. Garratt, and W.L. Physick, "The inland boundary layer at lowlatitudes. II: Sea-breeze influences", Bound-Layer Meteor., 33, 209,1985.

15. M. Segal, R.A. Pielke, R.W. Arritt, M.D. Moran, C-H. Yu, and D.Henderson, "Application of a mesoscale atmospheric modeling systemto the estimation of S02 concentrations from major elevated sources insouthern Florida", Atmos. Environ., 22, 1319, 1988.

16. R.T. McNider, M.D. Moran, and R.A. Pielke, "Influence of diurnal andinertial boundary layer oscillations on long-range dispersion", Atmos.Environ., 22, 2445, 1988.

17. D.G. Steyn, and I.G. McKendry, "Quantitative and qualitative evalua-tion of a three-dimensional mesoscale numerical model of a sea-breezein complex terrain", Mon. Wea. Rev., 10, 1914, 1988.

18. Z.J. Ye, M. Segal, J .R. Garratt, and R.A. Pielke, "On the impact ofcloudiness on the characteristics of nocturnal downslope flows", Bound-Layer Meteor., 49, 23, 1989.

19. I. Laszlo, H. Jacobowitz, and A. Gruber, "The relative merits of narrowband channels for estimating broadband albedos", J. Atmos. OceanicTech., 5, 757, 1988.

20. D.G. Bader and D.L. Ruschy, "Winter albedo characteristics at St. Paul,Minnesota", J. Appl. Meteor., 28, 227, 1989.

838

..,

l'~!e:~, .2-

~~D"

~'"><

2'.!.-~

~.!:

~i~E-n

=~. ~Q,

~

.., '". ~;;~

.::!~~Qi

:!~~~ n!:- " _.~~n:" ~

t ",..", ..-"'~ o!:=. ~ _.~j5.0. 8 "'-.~~><..

f~i.i c;:,- g S

~;::.8!,g,;e.l'i~. ; ..~"~~28 ~ ..~oe;"~g.

[;-:;2:?-;c6s

t.. &..."" o~-g.z- G .0

.~{.~

6£8

~::<-1

I/)'"'"II:~

e>-

e-'~8g~~~.....~

~(I)

';j'

i...9

gu..:~~ti:

¥..~.on

i

~

iIt' ... '" '"(..)

00'"0

~~~~:3QIe~

:0-r;:.-'"~00'"0

~i~od...:3QI

~"'I~I

od'

.,,:~"bO~

841

,,;:"

~