supplementary materials

SUPPLEMENTARY MATERIALS
Table S1 - Specifications of the flux components for the two set of modeled CO2 concentrations
used in the study.
Underlying flux
component
PCTM/GEOS4-CASA modeled CO2
concentrations
IFS-ORCHIDEE modeled CO2
concentrations
Biospheric
CASA GFEDv2 (Randerson et al. [1997])
ORCHIDEE (Krinner et al. [2005])
Oceanic
Air-sea CO2 exchange described by
monthly mean climatology (Takahashi et
al. [2002])
Air-sea CO2 exchange described by
monthly mean climatology (Takahashi et
al. [2002])
Anthropogenic
Seasonally variable global emissions
modeled using a 2-harmonic Fourier
series approximation (Erickson et al.
[2008])
EDGAR 3.0 1 × 1 global map for 1990
(Olivier and Berdowski [2001]), rescaled
to the CDIAC country level estimates for
1998
Wildfire/Biomass
burning emissions
Global Fire Emission Database version 2
(GFED2) (van der Werf et al. [2006])
Global Fire Emission Database version 2
(GFED2) (van der Werf et al. [2006])
Table S2- Aircraft (white) and TCCON (grey) sites used in the evaluation of the 4DVAR
analysis. Note that all the aircraft sites are located over North America.
Measurement
site code
Site name
Site latitude/ longitude
40.80 N, 97.18 W
Evaluation data
for January
2010
Y
Evaluation
data for June
2010
Y
BNE
Beaver Crossing, Nebraska
CAR
Briggsdale, Colorado
40.37 N, 104.30 W
Y
Y
CMA
Cape May, New Jersey
38.83 N, 74.32 W
Y
Y
DND
Dahlen, North Dakota
48.38 N, 99.00 W
-
Y
ESP
Estevan Point, B. Columbia
49.58 N, 126.37 W
Y
Y
ETL
East trout Lake, Saskatchewan
54.35 N, 104.98 W
Y
Y
HIL
Homer, Illinois
40.07 N, 87.91 W
Y
Y
LEF
Park Falls, Wisconsin
45.95 N, 90.27 W
Y
Y
NHA
Worcester, Massachusetts
42.95 N, 70.63 W
Y
Y
PFA
Poker Flats, Alaska
65.07 N, 147.29 W
Y
Y
SCA
Charleston, South Carolina
32.77 N, 79.55 W
Y
Y
SGP
36.80 N, 97.50 W
Y
Y
TGC
Southern Great Plains,
Oklahoma
Sinton, Texas
27.73 N, 96.86 W
Y
Y
THD
Trinidad Head, California
41.05 N, 124. 15 W
Y
Y
WBI
West Branch, Iowa
41.72 N, 91.35 W
Y
Y
BIA
Bialystok, Poland
53.23 N, 23.03 E
Y
Y
BRE
Bremen, Germany
53.10 N, 8.85 E
Y
Y
DAR
Darwin, Australia
12.42 S, 130.89 E
Y
-
EUR
Eureka, Canada
80.05 N, 86.42 W
-
-
GAR
Garmisch, Germany
47.48 N, 11.06 E
Y
Y
IZA
Izana, Tenerife
28.30 N, 16.50 W
-
Y
KAR
Karlsruhe, Germany
49.10 N, 8.44 E
-
Y
LAU
Lauder, New Zealand
45.04 S, 169.68 E
-
Y
LEF
Park Falls, Wisconsin
45.95 N, 90.27 W
Y
-
NYA
NY Alesund, Norway
78.92 N, 11.92 E
-
Y
ORL
Orleans, France
47.97 N, 2.11 E
-
Y
SGP
Lamont, Oklahoma
36.80 N, 97.50 W
Y
Y
SOD
Sodankyla, Finland
67.37 N, 26.63 E
-
Y
TSU
Tsukuba, japan
36.05 N, 140.12 E
Y
-
WOL
Wollongong, Australia
34.41 S, 150.88 E
Y
-
Figure S1- Inferred CO2 correlation length (3l) and variance (2) parameters for January (A
and C) and June (B and D). The covariance parameters are shown for the 15th day of the month,
1800h UTC at ~45 hPa (i.e., ~ 20.0 km).
Figure S2- CO2 concentrations from the 4D-VAR analyses based on the two background error
statistics –-statistics (A and B), NMC statistics (C and D), and the unconstrained model run (E
and F) for January and June 2010. These plots are shown for the 15th day of the month, 1800h
UTC at ~45 hPa (i.e., ~ 20.0 km).
Figure S3- Column-averaged CO2 mixing ratio analysis increments based on the two
background error statistics –-statistics (panel A), and NMC-based statistics (panel B), averaged
over the month of June 2010. As described in Fisher [2002], within the ECMWF 4D-VAR
system these analysis increments are calculated using the differences between the observations
and the model background (short forecast), the observation errors and the background errors.