Sin t

1
A new semi-empirical model of sea surface
microwave emissivity used to retrieve salinity,
wind speed and wave height
C. Gabarró1, J. Font1, A. Camps2, M.
Vall-llossera2, M. Portabella3 1Institut de
Ciències del Mar- CSIC. Barcelona. E-mail
cgabarro_at_icm.csic.es 2 Universitat Politècnica de
Catalunya. Dept. Teoria del Senyal i Comunicació.
Barcelona 3 KNMI Koninklijk Nederlands
Meteorologisch Instituut. De Bilt. Holland

• SMOS Soil Moisture and Ocean Salinity is an ESA
  mission to be launched on 2007. It carries an
  L-band polarimetric radiometer with multi-angular
  view capability.
• WISE (Wind and Salinity Experiment) field
  measurements were done to improve sea surface
  emissivity models at L-band for sea surface
  salinity (SSS) retrieval A radiometer and
  several buoys were installed in a fixed platform
  in the NW Mediterranean to measure TB, SST, SSS,
  WS and SWH.

• The main objective of this study is to retrieve
  SSS from WISE measurements (TB) taking the
  advantage of the multi-angular view.
• The possibility of retrieving the auxiliary data
  from the radiometric measurements themselves has
  been investigated.

III. Semi-Empirical Emissivity Models from WISE
Experiment
The polarised brightness temperature for each
incidence angle ? is modelled as the sum of the
theoretical model for flat surface emission and a
term dependent on surface roughness,
parameterised through wind speed (WS) and / or
significant wave height (SWH). Two
empirical models for ?TB_rough fitted from WISE
measurements were derived
The errors in retrieved SSS from WISE2001 data
using both models and in situ SST, WS and SWH
WS dep. SSS in situ - SSS ret 0.19
psu WS SWH dep. SSS in situ - SSS ret 0.13
psu

• dependent on Wind Speed 1

• dependent on WS SWH 2

Conclusion The WISE model with mixed dependence
provides the best salinity retrieval in this case.
IV. How to obtain the auxiliary data?
The auxiliary data needed to retrieve salinity
from rediometer measurements are WS, SWH and SST
among others. These data could be obtained from
external sources (meteorological and
oceanographic models, satellite measurements,
etc) or from the radiometer measurements
themselves. Here the last case is developed.
Wind Speed and Significant Wave Height data To
retrieve these parameters a least squares method
is used. The SSS, WS and SWH values that minimise
the distance between measured and modelled TB
represent the correct value for each parameter.
Looking at the plots of the cost function
dependent on SSS, WS and SWH (Figure 1) one can
realise that only one minimum is present, so it
is possible to retrieve these values without any
inconsistency. The cost functions with
restrictions consist on giving a priori knowledge
(reference) of the parameters as well as the
expected error of these a priori values (?2 ).
For these calculations HIRLAM and WAM model
outputs are used as reference data for WSref and
SWHref, respectively. A constant value is used
for SSSref and SST is supposed to be known with
exact value / an error of 1ºC. Two different
values for the standard deviations error of the
reference values are shown. The errors in
retrieved SSS, WS and SWH are ?2tb1,
?2SSS0.25, ?2WS4 and ?2swh0.25 ?2tb1,
?2SSS1.0, ?2WS6.25 and ?2swh1.0. SSS in
situ - SSS ret 0.20 / 0.22 psu SSS in
situ - SSS ret 0.28 / 0.32 psu WS in
situ - WS ret 1.03 / 1.07 m/s WS
in situ - WS ret 1.11 / 1.08 m/s SWH
in situ - SWH ret 0.28/ 0.28 m SWH
in situ - SWH ret 0.48/ 0.48 m Figure 2
compares the retrieved parameter (red), the in
situ measurement (green) and the model output
(blue) for each parameter for different
radiometric measurements. SSS retrieved with this
method is better than using directly WS and SWH
values from HIRLAM and WAM (SSS in situ - SSS
ret 0.34 psu).
COST FUNCTION
Figure 2
Figure 1
Sea Surface Temperature data TB is highly
non-linearly dependent on SST. Figure 3 shows the
distribution of the cost function without
restrictions. A direct and unique minimum is not
visible in this case. Figure 4 shows that
considering restrictions a clear minimum appears,
so it is possible to find it when the possible
solutions are restricted. The results after
retrieving SSS, WS, SWH and SST from radiometer
data appear to be a little bit better than for
the previous case ?2tb1, ?2SSS0.25, ?2WS4
, ?2swh0.25 ?2sst1.0 ?2tb1, ?2SSS1.0,
?2WS6.25 ?2swh1.0 ?2sst1.5 SSS in situ -
SSS ret 0.20 psu SSS in
situ - SSS ret 0.25 psu WS in situ -
WS ret 1.05 m/s WS in
situ - WS ret 1.13 m/s SWH in situ -
SWH ret 0.28 m SWH in situ
- SWH ret 0.48 m SST in situ - SSTret
0.20 ºC SST in situ -
SSTret 0.64 ºC
Figure 3
Figure 4
Conclusions Salinity is retrieved better when
WS, SWH and SST are determined from the
radiometric data than using the available
numerical models (WS, SWH) or poor quality data
(SST).
REFERENCES 1 A.Camps et al., L-Band Sea
Surface Emissivity Preliminary Results of the
WISE-2000 Campaign and its Application to
Salinity Retrieval in the SMOS Mission, Radio
Science,Vol. 38, No. 3, 2003. 2 C Gabarró et
al., A new empirical model of sea surface
microwave emissivity for salinity remote
sensing, Geophysical Research Letters, Vol. 31,
L01309, 2004
ACKNOWLEGMENT This study was funded by
ESA-ESTEC under WISE contract 14188/00/NL/DC and
by the Spanish National Program on Space Research
grants MIDAS ESP2001-4523-PE and ESP2002-11604-E.
The authors very much appreciate the cooperation
of Puertos del Estado for providing us with the
HIRLAM and WAM data for the periods required.