using space-borne measurements of hcho to test current understanding of tropical bvoc emissions paul...
TRANSCRIPT
Using space-borne measurements of HCHO to test
current understanding of tropical BVOC emissions
Paul Palmer University of Edinburgh
xweb.geos.ed.ac.uk/~ppalmer
Current model estimates show that tropical ecosystems represent 75% of global biogenic
NMVOC emissions
Guenth
er
et
al,
200
7
But how accurate are these estimates? How well do we understand observed surface flux variability?
Barkley et al, in prep., 2007
Because measurements are sparse including individual data points (and extrapolating them to plant functional
types) have a big effect on bottom-up models
Pfister et al, in review, 2007
From CH4
From isoprene
From other
Contribution of isoprene to Amazon chemical budget
NCAR MOZART-4 CTM
MODIS #1CL
MMODIS
#2
LAI/PFT maps
An integrative perspective is required
Net canopy VOC flux
Con
cen
trati
on
(z)
Column abundance
d[HCHO]/dt = [VOC][OH]k – [HCHO][OH]k’
In-canopy sinks
GOME HCHO columns: July 1998
[1016 molec cm-
2]
Biogenic emissionsPalmer et al, Abbot et al, Millet et al
Biomass burning*Columns fitted: 337-356nm
*Pixel: 320km x 40km * Fit uncertainty < continental signals * Only use cloud fraction<40%
Data
: c/o C
hance
et
al
South Atlantic Anomaly
Fu et al, Shim et
al
Curci et al
Palmer et al, Barkley
et al
Monthly mean AVHRR LAIMEGAN (isoprene)
Canopy model; Leaf age; LAI; Temperature; Fixed Base factors
GEIAMonoterpenes; MBO;Acetone; Methanol
MODEL BIOSPHERE
GEOS-Chem chemistry transport model
Chemistry and transport run at 2x2.5 degrees ANDsampled at GOME scenes
PAR, T
Emissions
Parameterized HCHO source from monoterpenes and MBO using the Master Chemical Mechanism
d[HCHO]/dt = [VOC][OH]k –[HCHO][OH]k’
GFED biomass burning
emissions
Master Chemical Mechanism yield calculations
Cu
mu
lati
ve H
CH
O y
ield
[p
er
C]
0 2 4 6 8 10 12 14 16 18 20 220.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
HC
HO
YIE
LD
PE
R C
RE
AC
TE
D
DAYS
NOX= 1 PPB NOX= 100 PPT
pinene
( pinene similar)DAYS
0.4
0 20 40 60 80 100 120 1400.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55C
umm
ula
tive
HC
HO
Yie
ld fr
om
iso
pren
e o
xid
atio
n (p
er C
)
TIME (HOURS)
NOX = 0.1 PPB
NOX =1 PPB
Figure 18. Formation of HCHO from isoprene. Vertical lines denote midnight of each day
Isoprene
HOURS
0.5NOx = 1 ppb
NOx = 0.1 ppb
Parameterization (1ST-order decay) of HCHO production from monoterpenes in global 3-D CTM – MAX 5-10% of column
Higher CH3COCH3 yield from monoterpene oxidation delayed (and smeared) HCHO production
Palmer et al, JGR, 2006.
C5H8+OH(i) RO2+NOHCHO, MVK, MACR
(ii) RO2+HO2ROOH
ROOH recycle RO and RO2
Month
ly
ATSR
Fire
counts
Sla
nt
Colu
mn
HC
HO
[1
016 m
ole
c cm
-2]
Day of Year
Significant pyrogenic HCHO source over South America
Good: Additional trace gas measurement of biomass
burning; effect can beidentified largely by
firecounts.
Bad: Observed HCHO is a mixture of
biogenic and pyrogenic – difficult to
separate without better temporal and
spatial resolution
ATSR Firecount
Remove HCHO if concurrent NO2 > 8x1015 molec/cm2
Barkley et al, in prep., 2007
Firecounts and GOME NO2 columns are used to remove pyrogenic HCHO signal over western South America
NO2 HCHO
10
15,
10
16
[mole
c/cm
2]
HCHO NO2
Model HCHO columns are typically 20% higher than GOME data
Model and observed
columns are better
correlated in the dry season
Ground-based and aircraft measurements of isoprene and/or HCHO are sparse but invaluable for evaluating
satellite dataTrostdorf et al, ACPD, 2007
Helmig et al, JGR, 1998
Kuhn et al, JGR, 2002
Kuhn et al, ACP, 2007
Kuhn et al, ACP, 2007
Barkley et al, in prep, 2007.
Helmig et al, JGR, 1998 MEGAN 2004
Ts
MEGAN 2004 T(1)
MEGAN 2006
Annual cycle of isoprene
Hypothesis: water availability has a role in determining the magnitude of isoprene emission in the dry season
In situ isoprene 2002
Tro
stdorf e
t al,
200
4
Isop
ren
e [
pp
b] Dry season
Trostdorf et al, ACPD, 2007
Carswell, et al, 2002 Huete et al, 2006
In situ isoprene 2002
Tro
stdorf e
t al,
20
04
Isopre
ne [
ppb]
Dry seasonLA
I
1999
Vegetation seasonal phenology (mean +/- sd). Satellite EVI and local tower GPP at Tapajos primary forest (km 67 site, 2002-2004).
Other factors affecting phenology?
Kuhn e
t al, 2
00
4
Dry season
Barkley et al, in prep, 2007.
GEOS-Chem(MEGAN) has only a weak annual cycle compared with data, symptomatic of model
deficiency
Bias = +102%; r2 = 0
Bias = +38%; r2 = -0.2
Bias = +180%; r2 = 0
Kuhn et al, JGR, 2002
Are bottom-up inventories biased towards dry season measurements?
GEOS-Chem over estimates surface [HCHO] during (1) the wet season and (2) night time
Model does NOT account for in-canopy chemistry and not a fair data comparison
HCHO Columns Over NW South America
Use GOME NO2 and ATSR firecounts to remove pyrogenic HCHO S
lan
t C
olu
mn H
CH
O [
10
16 m
ole
c cm
-2]
Month
2.5
2.0
1.5
1.0
0.5
0.0
Q: What’s driving this seasonal distribution of HCHO?
In situ isoprene 2002
Tro
stdorf e
t al,
20
04
Isopre
ne [
ppb]
Dry season
Relating HCHO Columns to VOC Emissions
VOC HCHOhours
OH
hours
h, OH
Local linear relationship between HCHO and E
kHCHO
EVOC = (kVOCYVOCHCHO)HCHO
___________
VOC source
Distance downwind
HCHO Isoprene
-pinenepropane
100 km
EVOC: HCHO from GEOS-CHEM CTM and MEGAN isoprene emission model
Palmer et al, JGR, 2003.
Net
LL-VOC ELL-VOC + SL-VOCESL VOC = HCHO
kHCHO
(kVOCYVOCHCHO)___________ =
Background due to CH4, CH3OH
, GEOS-Chem chemistry mechanism
Isoprene emission E [1013 atomC cm-2 s-1]
May
AugJul
Jun
r = 0.9
r = 0.9
r = 0.8
r = 0.9
Mod
el
HC
HO
[10
16 m
ole
c cm
-2]
Slope = 2000-2200 s
Intercept (background) = 5-6x1015 molec/cm2
Isoprene emissions [1013 molec/cm3/s]
MEGAN GOME
Apr
Jun
Aug
Oct
MODIS EVI
Bottom-up emission inventories typically represent within-canopy measurements:(1) Within-canopy turbulence and chemistry are sub-grid scale processes in global 3-D CTMs (2) Artificially increase [OH] to remove isoprene faster would be problematic in global CTMs
Con
cen
trati
on
(z)
Net canopy VOC flux
Column abundance
d[HCHO]/dt = [VOC][OH]k – [HCHO][OH]k’
In-canopy sinks
Provided GEOS-CHEM d[HCHO]/dt
is correct then canopy fluxes of VOCs inferred
from HCHO columns are more suitable for global
models
What we’ve shown….
Satellite observations of HCHO have strong (and distinct) pyrogenic and biogenic signatures.
GOME HCHO data are broadly consistent with the temporal variability observed by ground-based data, particularly the partitioning between wet and dry season.
GOME HCHO data are qualitatively consistent with bottom-up isoprene emissions in the dry season (when model bias is greatest).
Bottom-up models (here, we pick on MEGAN!) lack data to provide robust isoprene estimates over South America.
Isoprene emissions inferred from GOME represent the canopy-atmosphere flux – what global 3-D CTMs want.
Open questions that still need to be answered…
How do we reconcile the apparent discrepancy between ground-based measurements of isoprene flux and concentration and oxidation products?
Are GOME isoprene fluxes more consistent with ground-based data? [Calculations running as we speak]
Why are isoprene fluxes in the dry season higher than in the wet season? Light vs drought: are GOME isoprene fluxes more consistent with seasonal changes in EVI or drought indices?]
How important is isoprene to the regional carbon budget?
Will better spatial and temporal resolved satellite data improve estimates?
SPARE SLIDES
Vertical column retrievals
8 x 1016 molec cm-2T
ransm
issi
on
Chance et al, GRL, 2000
337-356 nm (O3, NO2, BrO, O2-O2)
1) Direct fit of observed radiances: slant columns
AMF = AMFG w() S() d1
0
Radiative transfer
Normalised HCHO profile
Palmer et al, JGR, 2001
2) Air-mass factor calculation: vertical columns
Estimated Error Budget
Slant column fitting: 4x1015 molec cm-2
AMF:
1) UV albedo (8%)
2) Model error (10%)
3) Clouds (20%)
4) Aerosols (20%)
Subtotal 30%
For a vertical column of 2x1016 molec cm-2 and AMF of 0.7
TOTAL = 9x1015 molec cm-2
Month of 2000
50
0
Mod
el b
ias
[%]