Determining emissions of environmentally important
gases using data from aircraft and satellites
with Dorian Abbot, Arlene Fiore, Colette Heald, Daniel Jacob, Dylan Jones, Jennifer Logan, Loretta Mickley, Bob Yantosca
Harvard UniversityRandall Martin, Kelly Chance, Thomas Kurosu
Harvard-SmithsonianGlen Sachse NASA Langley; Don Blake UCI;David Streets Argonne National Laboratory;
Henry Fuelberg, Chris Kiley FSU
http://www.people.fas.harvard.edu/~ppalmer
Paul Palmer
Global 3d chemistry transport
model
Top-down and bottom-up emission inventories
GOME,
MOPITT,
SCIAMACHY
TES, OMI
CMDL network for CO and CO2
CO inverse modeling•Product of incomplete combustion; main sink is OH
•Lifetime ~1-3 months
•Relative abundance of observations
•Previous studies found a discrepancy between Asian emission inventories and observations
RH + OH … CO
1000s km
Direct & indirect emissions
CMDL site
Many 100s km10s km
Increasing model transport error
Limitation of remote data for inverse model calculations
TRACE-P data can improve level of disaggregation of continental emissions
110 E 120 E 130 E 140 E 150 E 160 E
Longitude
0 N
10 N
20 N
30 N
40 N
50 N
Lat
itu
de
DC-8 FlightsP-3B Flights
cold front
cold air
warm air
Main transport processes:
DEEP CONVECTION
OROGRAPHIC LIFTING
FRONTAL LIFTING
100 E 130 E 160 E 190 E 220 E 250 E 280 E
Longitude
0 N
10 N
20 N
30 N
40 N
50 N
60 N
La
titu
de
DC-8 FlightsP-3B Flights
Feb – April 2001
Forward model (GEOS-CHEM)
Inverse model
P3B, DC8 observations y
Emissions x
FF
BB
BF
Modeling Overview
xs = xa + (KTSy-1K + Sa
-1)-1 KTSy-1(y – Kxa)
SS = (KTSy-1K + Sa
-1)-1
y = Kxa +
DACOM (Sachse)
TRACE-P CO Emissions Inventories Biomass
burning: Variability from observed daily firecount data (AVHRR)
(Heald/Logan)
Anthropogenic emissions for Y2K1 (fuel consumption)
(Streets)
Tagged model CO simulation for TRACE-P
ChinaJapan
Southeast Asia
Korea
Rest of World
[OH] from full-chemistry model (CH3CCl3 = 6.3 years)
Global 3D CTM 2x2.5 deg resolution
GEOS-CHEM
CO
[p
pb
]
Lat [deg]
Observation
A priori
A priori emissions have a large negative bias in the boundary layer
xs = xa + (KTSy-1K + Sa
-1)-1 KTSy-1(y – Kxa)
SS = (KTSy-1K + Sa
-1)-1
x = state vector (emissions)y = observation vector (TRACE-P CO, ppb)
Choice of state vector…-Aggregate anthropogenic emissions-Aggregate Korea/Japan
Inverse Model (a.k.a. Weighted linear least-
squares)
Gain matrix
o Emission uncertainties for Asia Sa :
Anthropogenic (D. Streets): China (78%), Japan (17%), Southeast Asia (100%), Korea (42%)
Biomass burning: 50%; Chemistry (largely CH4): 25%
o Observation uncertainty Sy :
Measurement accuracy (1%)
Representation (14ppb or 25%):
Model errors…
GEOS-CHEM
GEOS-CHEM
2x2.5 cell
TRACE-P
Error specification is crucial
Estimated: 1 sigma value about mean observed 2x2.5 value
All latitudes
(measured-model) /measured
Alt
itu
de [
km
]
Mean bias
RRE
Model error: (y*RRE)2 ~38ppb (>70% of total
observation error)
MODEL ERROR
Kore
a +
Jap
an
Sou
theast
Asia
Ch
ina (
BB
)
Ch
ina
(an
thro
pog
en
ic)
A prioriA posteriori
1-sigma uncertaint
y
Rest of World
Our best estimate is insensitive to inverse model assumptions
GEOS-CHEM
CO
[p
pb
]
Lat [deg]
Observation
A priori
A posteriori
A posteriori emissions improve agreement with observations
[1018 molec cm-2]
MOPITT shows low CO columns over Southeast Asia during TRACE-P
GEOS-CHEM
MOPITT
MOPITT – GEOS-CHEM
[1018 molec cm-2]c/o Heald, Emmons, Gille
Largest difference
Next steps with CO…
Multi-species inversion will bring additional information:
-CH3CN will bring information about biomass burning
-CO2 used to disaggregate emissions from Korea and Japan (CO2/CO)
Direct & indirect emissions
Can calculate emissions of anthropogenic halocarbon X given the X:CO slope and CO emissions
Western Pacific
CO, species with CO, +many other species
Asian continent
Blake group: CH3CCl3, CCl4, Halons 1211, 1301, 2402, CFCs 11, 12, 113, 114, 115
2 kmFresh
emissions
Back-trajectories of top 5% of observed values indicate local
sources
Proxy for OHOnly a strong local source
CO:CH3CCl3 relationships
= value above
“background”
0
5
10
15
20
25
30
35
40
45
Gg
/yr
CH 3CCl 3
CCl 4
CFC-11
CFC-12
CH3CCl3,CCl4,CFCs 11 & 12):
-represents >80% of East Asia ODP (70% of total global ODP)
-103.1 ODP Gg/yr (East Asia)
East Asia ODP = 70%
Global ODP = 20%
Eastern Asia estimates
Large global & regional implications
Methodology has the potential to monitor magnitude and trends of emissions of a wide range of environmentally important gases
Previous workThis work
0.9
1.4
2.3
3.0
Platform multiple ERS-2 Terra ENVISAT Space station
Aura TBD TBD
Sensor TOMS GOME MOPITT MODIS/MISR
SCIAMACHY MIPAS SAGE-3 TES OMI MLS CALIPSO OCO
Launch 1979 1995 1999 1999 2002 2002 2004 2004 2004 2004 2004 2005
O3 N N/L L L N/L N L
CO N N/L L N/L
CO2 N/L N
NO L
NO2 N N/L N
HNO3 L L
CH4 N/L N
HCHO N N/L N
SO2 N N/L N
BrO N N/L N
HCN L
aerosol N N N L N N
N = NadirL = Limb
Satellite data will become integral to the study of tropospheric chemistry in the next
decade
• Nadir-viewing SBUV instrument
• Pixel 320 x 40 km2
• 10.30 am cross-equator time
• Global coverage in 3 days
Global Ozone Monitoring Experiment
•HCHO slant columns fitted: 337-356nm
- fitting uncertainty 4 x 1015 molec cm-2HCHO JULY 1997
Isoprene
Biomass Burning
Isoprene dominates HCHO production over US during
summer Southern Oxidant Study 1995
North Atlantic Regional Experiment 1997
[ppb]
Surface source (mostly isoprene+OH)
Continental outflow
Alt
itu
de
[km
]
Alt
itu
de
[km
]
measurements GEOS-CHEM model
Defined background CH4 + OH
Using HCHO Columns to Map Isoprene Emissions
isoprene
HCHOhours
OH
hours
Displacement/smearing length scale 10-100 km
h, OH
EISOP = ___________kHCHO HCHO
Yield ISOPHCHO
[1016molec cm-2]
GEOS-CHEM GOME
r2 = 0.7 n = 756Bias = 11%
HCHO columns – July 1996HCHO columns – July 1996
Model:Observed HCHO columns
GEIA isoprene emissions
[1012 atoms C cm-2 s-
1]
BIOGENIC ISOPRENE IS THE MAIN SOURCE OF HCHO IN U.S. IN SUMMER
[1012 atom C cm-2 s-
1]
GOME isoprene emissions (July 1996) agree with surface measurements
r2 = 0.77
Bias -12%
50
GEOS-CHEMGOME GOME GEOS-CHEM
1016 molecules cm-2
SEASONAL VARIABILITY IN GOME HCHO COLUMNS
0 2.5
r>0.75bias~20%
MAR
APR AUG
MAY
JUN
SEP
JUL
OCT
•Interannual Variability ~30%
10
16 m
ole
cu
les c
m-2
°C
0
2.5
-2
2
GOME T GOME
95
INTERANNUAL VARIABILITY IN GOME HCHO COLUMNS (1995-2001)
August Monthly Means & Temperature AnomalyT
97
98
01
00
99
96