FIRST GLOBAL MEASUREMENT OF
MIDTROPOSPHERIC CO2 FROM NOAA POLAR
SATELLITES: TROPICAL ZONE
A. Chédin, S. Serrar, N. A. Scott, C. Crevoisier and R. Armante
NOAA CMDL CCGG Cooperative Air Sampling Network
TOVS CO2- channels selected
Radiative transfer model simulations from a tropical atmosphere data base
2
Wavelength,
µm
14.7
14.5
+ 0.002 ± 0.02
4
14.2
- 0.27 ± 0.03
5
14.0
- 0.32 ± 0.02
6
13.7
- 0.32 ± 0.02
0.1
pressure (mb)
1
∂qCO 2
– co2 +3%
1
10
100
1000
-0.01
0
0.01
0.02
0.03 0.04 0.05 0.06
normalized functions
0.07
0.08
ch. 2
ch. 3
ch. 4
ch. 5
ch. 6
MSU3 temperature weight function and
mean co2 channels weight function
cover ~ the same pressure range
10
100
1000
-0.06
co2 channels mean jacobian
MSU3 temperature weight function
+ 0.13 ± 0.02
3
Mean co2 jacobians ∂TB
0.1
CO2: +3%
pressure (mb)
Channel
-0.05
-0.04
-0.03
-0.02
delta Tb (K/km)
-0.01
0
0.01
0.09
Retrieval method of CO2 from NOAA polar satellites
¾ A non-linear regression approach : Multilayer Perceptron (Rumelhart,1986)
q
HIRS
and
MSU
co2 var
TB
–q
co2 var
qref =355 ppmv
ref
– TB TIGR
~ central value
for NOAA-10
2 hidden layers
¾ MLP trained on the TIGR data set with variable qco2 drawn at random (341Æ369 ppm)
¾ Noised TB (instrumental and model noises)
¾ 49 MLPs trained (6 surface elevations (1013 to 875 hPa) over land and one over sea,
and 7 viewing angles (from nadir to 40°))
Application of Neural Network to observations requires knowledge of systematic biases
between simulations and observations
(Simulations - Observations) systematic biases calculation
Radiosoundings collocated
with satellite observations
T(P), H2O(P), Ts
Fixed greenhouse gases concentrations
Radiative
Transfer
Model
Simulated
brightness
temperature
(CO2 = 355ppm, CO = 100ppb, CH4 =
1.8ppm, N2O = 308ppb)
Ozone profile climatology (UGAMP)
NOAA-10
July 1987 – July 1991
Observed
brightness
temperature
Mean (Calc. – Obs.)
Over the period
these biases allow connection between ‘simulations world’ and ‘observations world’
TBs in NN inputs are TBobs + Mean (calc. – Obs.)
Global Maps of Mid-to-High tropospheric CO2
Monthly - 15°X15° (1° moving average)
CO2 (ppmv)
Number of items averaged
Global Maps of Mid-to-High tropospheric CO2
Monthly - 15°X15° (1° moving average)
CO2 (ppmv)
Number of items averaged
Dispersion of CO2 retrievals
Global Maps 15°X15° (1° moving average)
Standard deviation seasonal variability
Month
Minima in summer maxima in spring
Stdv of the method (StdM) ~ 3 ppm
Is the natural variability the only cause
of the difference?
Stdv in ppmv
Dispersion of CO2 retrievals
CO2 in ppmv
Stdv in ppmv
higher Stdv often localised in regions of CO2 strong gradients
causes other than natural variability have been considered
(tropospheric ozone variation, aerosols)
Mean Seasonal Cycles
As seen by NOAA-10 (5-14 km; 1987-1991)
As measured in situ (8-13 km; 1993-1999)
Commercial aircrafts (Matsueda et al., 2002)
Time variations of the CO2 concentration
as seen by NOAA-10
362
0N-20N
CO2 retrievals - 5° zonal means
360
CO2 (ppmv)
358
356
354
352
350
348
346
8707
8807
8907
Month
9007
9107
360
20S-0N
358
CO2 (ppmv)
356
354
352
350
348
346
344
8707
8807
8907
9007
9107
Month
Solid line : sum of the long term trend
and four harmonics
CO2 Growth Rate as seen by NOAA-10
(1987-1991)
Mean Northern hemisphere 1.76 ppm/yr
Mean Southern hemisphere 1.80 ppm/yr
Values consistent with the one observed
at the surface = 1.75ppm/yr
(Conway et al., 1994)
CO2 and ENSO
As seen by NOAA-10
(Chédin et al., 2003) (1987-1991)
As measured in situ
(Matsueda et al., 2002) (1993-1999)
Southern Oscillation Index
4
NH (20N-0)
SH (20S-0)
3
Growth rate (GR)
time evolution
2.5
2
1.5
1
0.5
8707
8807
8907
9007
9107
4
Southern Oscillation Index
3
2
SOI time
variation
SOI
1
0
-1
-2
-3
-4
8707
8807
8907
9007
9107
Month
6
NH (20N-0)
SH (20S-0)
5
CO2 growth rate (ppmv/yr)
CO2 growth rate (ppmv/yr)
3.5
8801
Correlation
SOI/GR
4
3
2
1
0
-4.5
-3.5
-2.5
-1.5
-0.5
0.5
SOI
1.5
2.5
3.5
4.5
Conclusions and perspectives
The method used to infer CO2 from NOAA polar satellites has proven
its ability to retrieve important features of the distribution of CO2
and its time evolution (monthly, seasonaly and yearly) :
¾ Mean rate of rise of CO2 of 1.78 ppm/yr over NOAA-10 period
¾ Seasonal cycle and impact of ENSO in agreement with Matsueda
findings from in situ observations
And now ?
Analyze CO2 data in the tropics and elucidate factors influencing its
variation (sources, sinks, transport)
Extend the period to the 25 years of NOAA/TOVS observations
(1979-2003)