satellite-derived pm2.5 emissions from wildfires for …satellite-derived pm2.5 emissions from...
Post on 17-Aug-2020
5 Views
Preview:
TRANSCRIPT
Satellite-derived PM2.5 Emissions from Wildfires for Air Quality Forecast
Xiaoyang Zhang1,2, Shobha Kondragunta1, Felix Kogan1, Jerald D. Tarpley1, Wei Guo3, Christopher Schmidt4
May 18, 2006, New Orleans
1NOAA/NESDIS Center for Satellite Applications and Research; 2Earth Resources Technology, Inc; 3I.M. System Group, Inc; 4Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin
Outline1. Fuel loading dataset derived from MODIS data2. AVHRR NDVI-controlled combustion and
emission factors3. GOES fire sizes4. PM2.5 emissions across the continental USA5. Summary
Modeling Emissions from Biomass Burning
GOES fire size
Fuel type AVHRR moisture condition
MODIS vegetation properties
CMAQ model
Emissions
Fuel loading Fraction of fuel consumption
Emission FactorBurned area
Model of Biomass Burning Emissions
AMCFE =E---biomass burning emissions (kg)A---burned area (km2)M--biomass density/fuel loading (kg.km-2)C--fraction of combustionF--fraction of emission
Standard formula (Seiler and Crutzen, 1980):
(1)
i and j define the fire (pixel) locations l is the fuel typek is the time period
Spatial-temporal-distributed formula:
(2)∑∑∑∑= = = =
=K
k
L
l
J
j
I
iijklijklijkijkl FCMAE
1 1 1 1
Biomass (Live Fuel) Estimates from Generalized Allometric Models and MODIS Data
Shrubs GrassesForests
Broadleaf forests
Needleleafforests
Vegetation Continuous Field (MOD44B) Land Cover Type (MOD12Q1)
Percent tree
Percent non-tree vegetation
LAI (MOD15A2)
Foliage Biomass
Biomass Components
General Allometric
Models
MODIS Data
Mixed forests
Percent bare land
Savannas
Species-specific Allometric Models
Foliage-Based Allometric Model
11
βα fc MM =
Mc – Branch biomass or aboveground biomass in a tree (kg)
Mf -- foliage biomass
α1, β1 -- coefficients.
(3)
Foliage Biomass Derived from LAILeaf mass (Mf) (kg C/m2) is calculated using the formula (Heinsch et al., 2003):
Mf=LAI/SLA (4) SLA—the specific leaf area (projected leaf area Kg-1 leaf C) for a pixel, which is obtained from a look-up table. LAI--the leaf area index, obtaining from MODIS LAI product. Forest LAI value in sub-pixel is calculated using LAI product, land cover product, and percent tree product.
MODIS Land cover type SLAEvergreen Needleleaf
Forest (ENF)21.1
Evergreen Broadleaf Forest (EBF)
23.3
Deciduous NeedleafForest (DNF)
31
Deciduous Broadleaf Forest (DBF)
26.2
Mixed Forest (MF) 21.5Grassy woodland (WL) 33.8
Wooded Grassland (Wgrass)
33.8
Close shrub (Closed Shrubland)
12
Open shrub (Open Shrubland)
19
Grass (Grasslands) 40Crop (Croplands) 36
Maximum monthly MODIS LAI in 2002
low high
>0 <7
MODIS Land Cover
MODIS Percent Tree Cover
low high
1 100%
MODIS Percent Nontree Vegetation
low high
(shrubs, crops, and herbaceous)
0 100%
Foliage Biomass
Foliage-based Generalized Allometric Models
Biomass Components
Tree Biomass Components
A. Foliage biomass (tons/ha)B. Branch biomass (tons/ha)C. Aboveground biomass (tons/ha)
Zhang, X., and S. Kondragunta (2006), Estimating forest biomass in the USA using generalized allometric models and MODIS land products, Geophysical Research Letter, 33, L09402, doi:10.1029/2006GL025879.
Biomass in Shrubs and Grasses
Total Shrub biomass
Grass biomass
2001078.002161.009.1 ccs VVM +−=
(5)
cslscflfl VMVMM +=
Ml— litter or CWD density in a pixelMlf -- litter or CWD density (kg/m2) for forests changing with forest typesMls--average litter or CWD density (kg/m2) in shrubs and grasses Vcf -- tree coverVcs --the non-forest vegetation cover
Litter or CWD Estimated Using Land Cover Type and Percent Vegetation Cover
(6)
Fuel Loading from MODIS — Litter and coarse woody detritus
<0.1Mg/ha >24 Mg/ha
>17Mg/ha<0.5Mg/ha
CWD
Litter
Fuel Moisture Category Factor (MCF) for Determining Combustion and
Emission Factors
Moisture Category Derived from AVHRR Global Vegetation Index (4km)
Vegetation health condition (VCI) index (Felix Kogan, 1997) :
VCI=100(NDVI-NDVImin)/(NDVImax-NDVImin) -------(7)
NDVI—Normalized Difference Vegetation Index in a week
NDVImax and NDVImin in a pixel are the weekly maximum and minimum values from1985 to 2005
Temporal Variation in Moisture Category
0
20
40
60
80
100
3 45 87 12917121325529733917 59 10114318522726931135331 73 1151571992412833253 45 87 129171213255297339
Day of year
VC
I
0
1
2
3
4
5
Moi
stur
e ca
toge
ry
Needle leaf forests in the western USA Grass in the western USAShrubs in the western USA Broadleaf forests in the eastern USAMixed forests in the eastern USA
2002 200520042003
Moisture Category --derived from AVHRR VCI
Early January in 2002 Early July in 2002
Moisture Category Factor--mcf
Moisture category
Canopy Shrub Grass CWD
Very dry 0.33 0.25 0.125 0.08
Dry 0.5 0.33 0.25 0.12
Moderate 1 0.5 1 0.15
Moist 2 1 2 0.22
Wet 4 2 4 0.31
Very wet 5 4 5 0.75
From Fire Emission Production Simulator (FEPS)
Fuel Combustion FactorConsumed fractions of tree canopy, shrub, grass (Anderson et al.,
2004)LCL=100*(1-e-1)mcf (8)
LCL--percent of fuel loading consumed for fuel type canopy, shrub, and grass, respectively
mcf—moisture category factor
The combustion factor for litter was assumed to be 100%.
Combustion factor for CWD:Cw=0.6(0.31+(0.03*(0.31-mcf))) (9)
Anderson et al. 2004 Fire Emission Production Simulator (FEPS) User’s Guide (version1.0)
PM2.5 Emission Factor (Ib/ton)
WetPM2.5
ModeratePM2.5
DryPM2.5
Litter, w 7.9 7.9 7.9
Wood 1-3” 11.9 11.9 11.9
Wood >3” 22.5 18.3 16.2Herbs and
shrubs 21.3 21.3 21.3
Canopy 21.3 21.3 21.3
Anderson et al. 2004 Fire Emission Production Simulator (FEPS) User’s Guide (version1.0)
GOES Fire Size
Fire occurrences detected from GOES satellite in 2002
•Spatial resolution: 4km•Temporal resolution: 30min• Instantaneous fire sizes in subpixels detected from 3.9 µm and 10.7 µm infrared bands
Half-hourly PM2.5 Emissions
Annual PM2.5 Emissions
Variation in GOES Fire and PM2.5 Emission with Land Cover Type
PM2.5 Emissions in Each State
0
5000
10000
15000
20000
25000
30000
35000
40000
Washington
IdahoM
ontanaM
ainem
innesotaN
. Dakota
Oregon
N.H
Michigan
V.T
Wiscons
New
York
S. D
akotaM
ichinganW
yoming
MA
CaliforniaC
onnN
evadaP
ENN
Iowa
Utah
NebraskaNew
JerseyO
hioIllinoisIndianaC
oloradoW
Virginia
MD
DEL
VirGinia
Missouri
Kansas
Kentucky
Arizona
N C
arolinaN
ewTennesseeO
klahoma
TexasA
rkansasS
Carolina
GeorG
iaAlabam
aM
ississippiLouisianaFloridaR.I.
PM
2.5
(ton)
0
200
400
600
800
1000
1200
1400
1600
1800
2000
Fire
siz
e (k
m2
PM2.5--2002PM2.5--2003PM2.5--2004Fire-size-2002Fire-size-2003Fire-size-2004
Comparison of Daily Emissions--April-December 2002
PM2.5 in 2002
0
5000
10000
15000
20000
25000
99 109
119
129
139
149
159
169
179
189
199
209
219
229
239
249
259
269
279
289
299
309
319
329
339
349
359
Day of year
Pm
2.5
(tons
National-Wildfire-Emissions-Inventory NFDRS fuelNESDIS fuelFCCS fuel
Near Real Time Emission(May 15, 2006)
Summary• MODIS data combined with foliage-based allometric
models provide a robust tool to establish fuel loading data over a large domain. This dataset is easy to update.
• PM2.5 emissions vary greatly with ecosystems, state, season, and year.
• Instantaneous fire size from GOES satellite is not the same as burned areas. Effort is needed to accurately retrieve burned areas from satellite-based fire sizes.
• Future work will derive reasonable burnt areas by investigating GOES fire size, MODIS fire counts, AVHRR fire counts, and national inventory burned areas in spatial and temporal patterns.
THANK YOU
Litter and Coarse Woody Detritus(CWD)
MODIS Land cover type Litter (kg/m2) CWD pool (kg/m2)
Evergreen Needleleaf Forest (ENF) 2.175 3.14Evergreen Broadleaf Forest (EBF) 1.5 1.76Deciduous Needleaf Forest (DNF) 0.85 1.6Deciduous Broadleaf Forest (DBF) 0.85 1.6
Mixed Forest (MF) 1.67 2.2Grassy woodland (WL) 0.63 0.176
Wooded Grassland (Wgrass) 0.63 0.176Cshrub (Closed Shrubland) 1.24 1.21Oshrub (Open Shrubland) 1.24 1.21
Grass (Grasslands) 0.3625 0.06Crop (Croplands) 0.3625 0.06
Non forest vegetation (average from shrub and grass)
0.5019 0.1657
Litter pool and CWD pool modified from Matthews et al. (1997).
Fire size comparison between GOES data and NEI in 2002Matchup comparison (10' grid)
05
101520253035404550
0 200 400 600 800 1000 1200Fire size from NEI (km2)
Fire
siz
e fro
m G
OES
(km
2)
Matchup comparison (20' grid)
0
50
100
150
200
250
0 500 1000 1500 2000Fire size from NEI (km2)
Fire
siz
e fro
m G
OE
S(k
m2)
Machup comparison (30' grid)
0
50
100
150
200
250
300
350
0 500 1000 1500 2000Fire size from NEI (km2)
Fire
siz
e fro
m G
OE
S(k
m2)
Machup comparison (40' grid)
0
50
100
150
200
250
300
0 500 1000 1500 2000Fire size from NEI (km2)
Fire
siz
e fro
m G
OES
(km
2)
Matchup comparison (60' grid)
0
100
200
300
400
500
0 500 1000 1500 2000Fire size from NEI (km2)
Fire
size fro
m G
OES
(km2)
050
100150200
250300350400450
0 50 100 150 200 250 300 350 400 450
Inventory Burned Area (km2)
GO
ES F
ire S
ize
(km
2 Whole CONUS
top related