501
360
1.0
Figure 5 - As for Figure 4 except for 90 GHz and h-Q approach.
As shown by Figure 4, it seems impossible to retrieve temperature (simultaneously with soil moisture)
with 23.8-36.5 GHz channels when soil moisture is high. Apart from this case, temperature and soil moisture can
be retrieved with a rms error of about 4 K and 0.04, respectively.
4 - SATELLITE DATA ANALYSIS
Methods established from the PORTOS-93 experiment can be applied to SSM/I data. Preliminary results are
discussed here for, up to now, METEOSAT-derived temperatures are only available for two days: june 3 and 23.
As explained in section 2.3.3, Champeaux and Jullien (1993) produced a vegetation map of the EFEDA domain
from which SSM/I homogeneously vegetated pixels could be selected. Table 2 presents the 14 SSM/I chosen
pixels.
Target Location NDVI class Prescribed Height June-23-NDVI and Night-IR-Emissivity
1
39°37N-2°22W
Summer crops
1 m
0.165
0.819
2
39°07N-2°07W
Summer crops
1 m
0.150
0.870
3
40°37N-5°22W
Grass & Shrubs
1 m
0324
0.891
4
38°22N-4°37W
Grass & Shrubs
1 m
0.292
0.922
5
37°53N-4°07W
Olive groves
3 m
0.159
0.942
6
39°37N-0°37W
Olive groves
3 m
0.232
0.875
7
39°22N-2°53W
Vineyard
1 m
0.172
0.958
8
38°37N-1°22W
Vineyard
1 m
0.117
0.906
9
40°22N-6°22W
Forests
5 m
0.452
0.867
10
39°37N-5°22W
Forests
5 m
0.390
11
38°53N-5°22W
Dry crops
1 m
0.136
12
39°53N-5°07W
Dry crops
1 m
0.192
0.860
13
41°53N-5°37W
Spring crops
1 m
14
37°53N-4°37W
Spring crops
1 m
0.269
0.915
Table 2 - SSM/I considered pixels with corresponding NDVI-derived class. Average NDVI and
nighttime METEOSAT emissivity on June-23 are also displayed.
In Table 2, target-7 corresponds to the Tomelloso test-site. Vegetation height is prescribed according to
Bolle et al. (1993) vegetation description. This parameter is necessary if the continuous approach is to be used.
The most difficult step in the retrieval process is the soil-vegetation structure parameters determination. In that
view, IR-derived surface temperatures are particularly useful as explained below.
