Fig. 1 The study area and test fields.
Finally, the following data set was obtained as the ground
support for the remote sensing observations:
.
The results of soil temperature and soil moisture
monitoring of the test fields. All the temperature
measurements and the sampling for soil moisture de-
termination, have been done simultaneously with the
acquisition of the remotely sensed data. On the test
field A a special system was installed for the auto-
matic soil temperature measurements. Based on the
two full cycles of these measurements the dynamic of
the diurnal temperature changes of the soil has been
evaluated. The water content of the soil samples was
assigned by weighing method.
The terrestrial thermal imagery of the selected part of
the test field A has been taken, mainly for the Bidirec-
tional Radiation Distribution Function (BRDF) as-
sessment in the thermal infrared region of the elec-
tromagnetic spectrum. For this purpose the thermovi-
sion camera was placed in the middle of the circle and
the several thermograms have been registered looking
towards and back of the incidence radiation. The se-
quence of the 18-th imageries were recorded on each
of 20 azimuth direction (chosen thermogramms are
shown on Fig.13) . As the results of the detailed
analysis of remotely sensed data and in situ meas-
urements, the field B has been selected as „sample
area” for preprocessing the whole data set.
4. METHODOLOGY AND RESULTS
Thermal inertia modeling was performed for all of the test
fields.
Panchromatic photos, in average scales (1:9000, 1:3000),
was digitized with 600 dpi resolution, so 1 pixel on the
image corresponds to 0.38 and 0.13 m on the terrain,
Thermal radiation was digital recorded as an image of 8
columns and 90 rows, 1 pixel corresponds to 1.5 m.
Panchromatic and thermal images were initially pre.
processed for adjustment to soil albedo and temperature
distributions. The test fields are composed with loess and
loess-like soils. On the base of the references and previ.
ous own research the albedo for dry loess was assumed
as 0.33 and for wet 0.22. Thermal images was calibrated
using ground temperature measurements. The tempera-
ture was changing, for sample area (field B), from 16 to
20. C.
The thermal inertia modeling was performed on the pre
processed remote sensing data considering meteorologi-
cal conditions and geographical co-ordinates as follows:
— latitude of the center of the test field: 50? N, longitude:
20* E,
— Sun declination: +16° , inclination: -0.8°,
— maximum diurnal temperature differences of the soil
surface: 20 °C,
— range of the air temperature: 20 °C,
— average air temperature: 8 °C,
— average wind velocity: 1.1 m/s.
Image processing was carried out using the own com-
puter software basing of the Pratt's et. al.(1981) theoreti-
cal background.
The last step was the finale soil moisture calculation.
Remote sensed data and in sifu measurement collected
for the test fields were statistically analyzed. Relationship
between the following variables were considered in all
possible combinations:
e soil moisture (my - hillocks of micro-relief, my - hollows
of micro-relief-, ms - soil surface),
e thermal inertia [TI - calculated on the base of Pratts.
et. al. (1981) model, Tl, - calculated from the simple
equation: (1-A)/AT],
e albedo - A,
e maximum diurnal temperature differences - AT.
Variance-covariance matrix for the test field B is shown
bellow (Tab.2.)
me mu ms TI TI, A AT
1 0,62 | -0,53
0,45 0,87 | -0,78
0,13 | -0
0 -0,92 89
1 7)
63
I
Tab.2. Variance-covariance matrix.
The relationship between albedo, maximum diurnal tem-
perature differences, thermal inertia and soil surface is
presented on Fig.3.
The analysis of the results shown in Tab.2. and on Fig.
allow to state that:
282
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
