×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Mesures physiques et signatures en télédétection

59
2.2. Water vapor dependency
2.2.1. Comparison of the Mid-Latitude and Tropical retrieved temperatures for land situation. Aset of
coincident ATSR brightness temperatures and surface temperatures was provided us by Gu (Crau92
experiment, Gu, 1993). The ground surface temperatures were measured over La Crau, an arid plain in
South-Eastern France, with a thermal camera. The principal interest of the soil radiometric measurements
is that both emissivity and atmospheric reflected radiances are included within. Thus these data can be
directly compared with the retrieved surface temperatures with e=l.
Fig. 3 shows the comparison between the ground truth temperatures and the Split-Window
retrieved temperatures with different sets of coefficients : the SW retrieved temperatures vary greatly
with the set of coefficients chosen (differences can reach 6 K for some cases). The Tropical retrieved
temperatures are always greater than the Mid-Latitude ones, the All-Situation (1761) ones being between
the two. But in all cases the true surface brightness temperatures always proves to be between the
different SW retrieved temperatures.
Figure 3. Comparison between true Ts and retrieved Tsw
with mid-latitude, tropical and all situations coefficients.
To conclude, the differences between the SW retrieved temperatures with the different sets of
coefficients must be taken into consideration when considering land pixels. This lead us to take care
about the set of coefficients used, and more generally to give up the idea of world-wide Split-Window
coefficients. Consequently we introduced a dependency with the total water vapor content into the
method. In practice, we have ranged the 1761 situations into 8 classes of mean water vapor amount
(instead of 3 classes of atmospheric types), and constructed in this way 8 sets of water vapor dependant
SW coefficients.
2.2.2. The 8 Split-Window water vapor classes. The mean water vapor amounts W of the 8 classes are
0.26 g/cm^, 0.72 g/cm^, 1.26 g/cm^, 1.71 g/cm^, 2.21 g/cm^, 2.76 g/cm^, 3.44 g/cm^ , and 5.01 g/cm^.
For each class (about 100 to 300 situations) the Split-Window coefficients have been computed. Fig. 4
shows the dependency of the coefficients with the water vapor content for a given emissivity (e=1). The
aj i and ai2 coefficients actually show a dependency with W, while the variation of c seems more erratic.
23. Applications and results
2.3.1. SST retrieval on the English Channel with the Split-Window technique. This study has been
done on an ATSR image over the English Channel with a coincident radiosounding. After the application
of a cloud mask 12000 cloud-free sea pixels remained. The radiative transfer model Lowtran7 (Kneizys
et al., 1989) combined with the local radiosounding has been used to calculate the reference SST (with
e= 0.99, from Masuda et al. (1988) for nadir views).
The Mid-Latitude coefficients were first applied to retrieve SST : in that case the mean error is
0.3 K (and this error reaches 0.98 K when using the Tropical coefficients).