a
ht,
ant
ric
ing
nts
to
ant
on.
nts
ere
to
1ce
ind
cal
xt,
om
ind
) is
de Jong, Steven
4 STATISTIC AND GEOSTASTIC ANALYSIS
4.1 Statistical analysis of the survey data
One of the objectives of this study is to gain insight in the spatial distribution of surface temperature in the area and
whether significant differences exist between the surface temperatures of the land cover types. This question can be
answered with the temperature data derived from the DAIS image. For each land cover type, the surface temperature is
derived by digitizing polygons on the DAIS thermal image and sampling the temperature values. These temperature
values and their statistical properties are then analysed. Figure 2 displays the difference in mean temperature per land
cover type.
50
4 ———
40 3 €
| al.
= "| x
Lu
E —
Be Ra
j
1
+
=
G
o
= 7 ; v Y : ; : Y
N= 294 25 70 41 154 33 100
vineyard grassland maquis garrigue
bare soil pine forest water
Land Cover
Figure 2: Average temperatures and standard deviations for each land cover type (derived from DAIS).
4.2 Geostatistical analysis of the field data
The three measured soil parameters: soil moisture content, porosity and bulk density are input parameters for the surface
temperature model. The purpose of this model is to gain information on the spatial distribution of the surface temperature in
time. In order to calculate the spatial distribution, continues maps of the soil parameters are required. These continues maps
are derived by interpolation of the point samples with the conditional simulation interpolation technique. The main
advantages of this technique are that insight in the prediction variance for each pixel (spatial) is obtained, as well as insight
in the model’s response to the input range.
For each of the variables, two variograms have been computed: one variogram for the cultivated (southern) area and the
other for the uncultivated (northern) area. Exactly 500 realisations have been computed with the conditional simulation
technique. Resulting maps of the mean value and the standard deviation for soil moisture content are given in figure 3. The
mean gives a good impression of the spatial distribution of the variable under consideration. The standard deviation gives an
impression of the range of possible results.
Legend
Standard
Legend Deviation of the
Se meine soil moisture
; fraction.
Figure 3: Regional distribution of A soil moisture content and B standard deviation of the soil moisture content
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 351
