4.2 Discussion
The results show that for soil which has a horizontal expanse over vast
areas VH polarization is the obvious choice for modelling, but it gets
attenuated when the soil texture is rough and there is sprouting of
crop. Hence a combination of both VV and VH backscatter parameters from
SAR satellite data were used for modelling. For soil salinity indices
development from Sentinel-2 data, Band 3 (Green) and Band 11 (Short Wave
InfraRed- SWIR) were used since saline soils reflect more in Green to
SWIR Bands. Hence NDSI generated was (refer to Section 3.1) used as
another parameter. The other parameters came from field data such as
temperature, soil moisture and soil electrical conductivity for surface
as well as root zone. Since satellite data is indicative of only surface
phenomenon hence a correlation was established between surface
parameters from field and root zone parameters. Thereafter, a
correlation was established between satellite and root zone parameters
collected from field by digging up to 2 ft (0.6 m) depth since crop
roots do not extent any further beyond. The Root zone parameters were
then used for modelling since all the above-mentioned correlative plots
gave an average R2-statistics in correlation above 0.8
(80%). The model gave highly accurate results in the limited resources
available with two polarization channels of SAR data and no complex
modelling approach and sophisticated chemical lab tests.
The estimated sub-surface soil salinity map in terms of Electrical
Conductivity for Rupnagar is shown in Figure 9-