SMOS measurements over the ocea

Download Report

Transcript SMOS measurements over the ocea

Optimization of L-band sea surface emissivity
models deduced from SMOS data
X. Yin(1), J. Boutin(1), N. Martin(1), P. Spurgeon(2)
(1)
(2)
LOCEAN, Paris, France
ARGANS, Plymouth, UK
Two scale + foam
Adjustment of some parameters of roughness and foam modeling
0
.2
2
5
lo
g
/2
)
1
0(k
2
Roughness:

1
.2
5
k
u

3
*
( )a
k
Omnidirectional wave spectrum Durden & Vesecki,1985 : Sk

0
a0? Original publication: a0=0.004; DV2, a0=0.008
 g
*

Th_30°
DV2
Nadir
~0.2K/m/s
At 15°C, a 0.1K Tb variation can be
generated by :
-0.2pss SSS variation
or
- 0.5m/s wind speed variation
Dinnat et al., IJRS, 2002, Radio Science, 2003
10m equivalent neutral wind speed (m/s)
c
Foam coverage (from Monahan & O'Muircheartaigh 1986):F

b
U
e
x
p
(

0
.
0
8
6
1)

T
1
0
b? c? original publication: b=1.95×10-5, c=2.55 ; ΔT =Tsea-Tair (neglected in this
first step study); in first SMOS SSS1 processing, F=0: no foam.
Foam emissivity (Stogryn, 1972): assumed to be correct
Modeling of SMOS Tbs
• Tb = Tbatm↑ + Rsea (Tbatm↓ + Tbsky) exp(-atm) + Tbsea exp(-atm)
Atmosphere
Tbsea= (Tbflat+Tbrough) (1-F) + F Tbfoam
Tbsea=esea SST
Ocean
esea derived from SMOS Tbs after correcting for all other effects
SMOS data used in the fit
SMOS Tbs: Tbs along track (~ no mixing of polarization) in the AFFOV
(good radiometric accuracy) from 19 ascending orbits in August (low
galactic noise) in the South Pacific (far from land) from 50°S to 0°N –
Incidence angles from 20° to 55°
Radiometric accuracy
5K
0
Along track in the AFFOV
5K
0
Incidence angle
(°)
Three different sets of wind induced components deduced from SMOS
1. There are totally 237501 samples in H polarization and 238469 in V polarization
collocated with ECMWF WS in range of 3-17ms-1.
2. 56% of the ECMWF wind speeds (136861 samples in H polarization and 137471 in V
polarization) have been collocated with SSMI WS in range of 3-17 ms-1 : +-0.5h +-50km
3. 126662 samples in H polarization and 127191 in V polarization, when the differences
between ECMWF and SSMI WS were restricted to be less than 2 ms-1
Data fitting
Wind induced component of emissivity deduced
(20-55° in step of 5°)
Er_SMOS (θi, p, ws)= Eres (θi, p) + Espectrum(a0; θi, p, ws)
3m/s<U<7m/s
a0 (prior=0.004 – 0.008)
8m/s<U<17m/s
b, c
Incidence angle (°)
a0
b
c
M1
0.0050
2.42×10-8
4.86
M2
0.0062
2.20×10-9
5.67
M3
0.0070
2.90×10-9
5.51
M
1
ECMWF
M2
ECMWF+SSMI
M3
SSMI
ECMWF
Results in
AFFOV
H pol. 20°
V pol. 20°
H pol. 30°
V pol. 30°
H pol. 40°
V pol. 40°
H pol. 50°
V pol. 50°
H pol. 55°
ECMWF WS
V pol. 55°
ECMWF WS
Results in EAFFOV ?
H pol. 0°
V pol. 0°
V pol. 10°
V pol. 20°
Monthly averages,
Pacific Ocean,
August 2010
SSS North-South profile,
Old model 1 (DV2)
Comparisons w.r.t WOA05
New parametrization for
roughness and foam
coverage
1. The tropical Southern Pacific ocean (20°S10°S- 140°W110°W) far away from continent
and island characterized by relative stable moderate wind speed and high SST; mean (standard
deviation) of SST and SSS are 24.5 (1.0) °C and 36.2 (0.3) pss
2. The high latitude Southern Pacific ocean (50°S45°S- 180°W100°W) characterized by
very variable wind speed and low SST; mean (standard deviation) of SST and SSS are 9,8 (1.8)
°C and 34,4 (0.2) pss
SMOS SSS retrieved with
the pre-launch model 1
SMOS SSS with the
new model M1
in red for the tropical Southern Pacific and in
green for the high latitude Southern Pacific
Summary
• SMOS data evidence that Tb(U) is non linear
• A reasonnable fit to SMOS data is obtained when introducing a foam
coverage parametrization close to Monahan and Muircheartaigh (1986),
(this foam coverage may be peculiar to L-band and depends on the foam
emissivity model)
• Parameter for the wave spectrum (a0) slightly higher than 0.004
• Preliminary validation shows improvement in mean retrieved SSS
• A larger set of SMOS data should be used for validating and/or improving
model
• Study the quality of SSS retrieved at high wind speed when putting a larger
error on ECMWF wind speed.
• Check SMOS measurements taken in the EAFFOV
Latitudinal drift ?
EH_SMOS - EH_Model, 0deg
EV_SMOS - EV_Model, 0deg
Statistics of SMOS SSS collocated with ARGO at +/-5days and +/-50km
during August ascending passes
Zone
Latitude
Longitude
SST (°C)
σSST (°C)
SSS (pss)
σSSS (pss)
Southern Pacific
50S-40S
20S-10S
180W-100W
140W-110W
9.8
24.5
1.8
1.0
34.4
36.2
0.2
0.3
pre-launch model 1
Wind speed (ms-1)
3-12
12-20
3-12
No. of collocations
11714
4060
4556
mean(SSSsmos-SSSargo) (pss)
0.08499 -0.89729
0.08477
σ (SSSsmos-SSSargo) (pss)
1.06832 1.37645
0.62445
median(SSSsmos-SSSargo) (pss)
0.1
-0.82515
0.0905
New model M1
No. of collocations
11718
4150
4551
mean(SSSsmos-SSSargo) (pss)
-0.04846 -0.40917
-0.06231
σ (SSSsmos-SSSargo) (pss)
1.0703 1.50764
0.63529
median(SSSsmos-SSSargo) (pss)
-0.0268 -0.5132
-0.055