3.1 LULC classification
The False Colour Composite (FCC) imageries were prepared for the 1990 LANDSAT satellite data using the band combination of “RGB 432” (Figure. 3a). FCC for the year 2000 LANDSAT ETM+ satellite data using band combination of “RGB 432” (Figure. 3b). For the 2010 LANDSAT satellite data, FCC band combination is “RGB 764” (Figure. 3c) while the band combination for the LANDSAT 8 satellite image for 2019 (Figure. 3d). After the preparation of FCCs and visual interpretation, LULC maps of the study region were prepared by using the unsupervised classification for 1990, 2000. 2010 and 2019 as shown in Figure. 4a, Figure. 4b, Figure. 4c, and Figure. 4d using the ISODATA technique.
After unsupervised classification, in which insights were gained about spectral variability in class, supervised classification of FCC images both the temporal database of LANDSAT images was carried out for 1990, 2000, 2010, and 2019 as given in Figure. 5a, Figure. 5b, Figure. 5c, and Figure. 5d using the Maximum Likelihood Classification.
3.2 Change detection analysis
To analyze changes for the period of past 29 years, supervised images of 1990, 2000, 2010, and 2019 were used as input images and the change detection of the present study region is shown in Table 2 in terms of Km2 and percentage. The supervised images for 1990, 2000, 2010, and 2019 were shown in Figures 5a, 5b, 5c, and 5d respectively. LULC change detection is classified into six categories namely vegetation, fallow lands, degraded lands, desertified lands, water body, and built-up lands. Usually, vegetation / agricultural texture is red color in the satellite imageries. For this LULC mapping, four dissimilar LANDSAT imageries (1990, 2000, 2010, and 2019) were collected and adopted to supervised classification and done mapping with the given color to red. In the year 1990 vegetation is 159.12 km2 (52.03%) is noticed. When coming to the year 2000 vegetation is decreased to 118.51 km2 (38.75%) and in the years 2010 and 2019 it is also decreased to 109.90 km2 (35.93%), 37.38 km2 (28.57 %) respectively. Fallow land is in olive color in the supervised images. In the year 1990 fallow land is noticed as 41.93 km2 (13.71 %) and in the year 2000, it is increased to 73.96 km2 (24.18 %). In the year 2010, it is decreased to 70.92 km2 (23.19 %) and in the year 2019, it is increased to 80.60 km2 (26.35 %). Degraded land is also called as barren land in the LULC classification. Here degraded land is noticed in the year 1990 as 56.79 km2 (18.57 %) and it is gradually increased for the years 2000, 2010, and 2019 to 62.93km2 (20.57%), 65.94 km2 (21.56%), and 68.25 km2 (22.31%) respectively. Coming to the desertified lands, active geomorphic changes take place in this study region. the wind is the geomorphological agent in the transportation or enhancement of dunes in the study part. Dune migration leads to an ecosystem imbalance in the study part. This leads the study part to be faced with desertification conditions and finally to be desertified. The yellow color is given to the desertified lands category in the supervised image. In the year 1990 it is noticed as 8.80 km2 (2.87 %) and in the year 2000, 2010, and 2019 it is continuously increased to 40.32 km2 (13.18 %), 44.91 km2 (14.68 %), and 51.99 km2 (17.00 %). Waterbodies are given to blue color in the supervised image. In the year 1990 waterbodies are noted as 0.89 km2 (0.29 %) and in the year 2000, 2010, and 2019 it is positively increased to 0.97 km2 (0.32 %), 1.02 km2 (0.33 %), and 1.41 km2 (0.46 %) respectively. Built-up land is given as green in color in the supervised imagery. In the year 1990 it is noticed as 49.31 km2 (14.48 %) and it is gradually increased for the years 2000, 2010, and 2019 as 9.17 km2 (3.00 %), 13.17km2 (4.31 %), and 16.23 km2 (5.31 %) respectively.
Table 3 labels the change detection in the total area (km2 and percentage) covered by different LULC between the years 1990 and 2019 procured from LANDSAT satellite data. The area under waterbody is slightly increased to 0.52 km2 (0.17 %), and the Built-up land is extended to 7.43 km2 (2.42 %). Vegetation is decreased to 71.76 km2 (23.46 %). Fallow land, Degraded lands, and Desertified lands showed increasing in trend to 38.67 km2 (12.64%), 11.46 km2 (3.74%), and 13.68 km2 (4.47%) (Fig 5b).
3.3 Soil Parameters analysis
Spatial erraticism of specific soil properties in the study region was predicted by digital soil mapping, these maps were produced in digital format in a rapid, efficient, operative, and little cost method. LD and desertification were shown as the spatial distribution of all the eleven in quantifiable terms via IDW interpolation methods using ArcGIS software-based statistical analysis tools (Sheng, 2010). Grounded on the pH values, a soil map for sampling sites was composed of the software ArcGIS 10.4 (Fig 6a). similarly, based on EC, SOM, N, P, K, Zn, Mn, Fe, Cu, and S values, soil maps were self-possessed as shown in Fig. 6b, c, d, e, f, g, h, I, j, and k, correspondingly, portraying the severity of LD and desertification in terms of soil parameters.
Areas like Nemakallu, Unthakallu, Uddehal, Bommanahal, Kuruvalli, of the North-Western (NW) part of the selected part exhibited a huge proportion of soils that were alkali or sodic, South-Eastern (SE) areas like Govindavada, Honnuru were calcareous or saline in nature, Bollanaguddam, Kalludevanahalli, Kuruvalli areas of north-eastern parts (NE) shows slightly to saline in nature. There is no acidic sample were traced in the study region (Fig 6a). Grounded on the EC standards, it has detected from the map (Fig .6b) that in the selected part, the maximum study part occupied by somewhat a little saline, and around of the south-western (SW) regions like Elanji and Kodaganahalli remained abstemiously saline. Founded on SOM ranges, it has detected from the map area (Fig .6c) that in the study, the maximum study area was high SOM likely 0.45-0.53. Founded on the NPK values, it is observed from the maps (fig .6d, 6e, 6f) that in the study area, deficiency of N is traced in NW and NE parts, deficiency of P is traced in SW part and moderate values in EW part, and lower in K values throughout the study region. Based on Zn values, it has observed from the map (Fig .6g) lower in the NE part and higher in the SW part. Founded on Mn values, it has observed from the map (Fig .6h) moderate throughout the region and NE part having variation in their values to moderate to high, based on Fe values, it has observed from the map (Fig. 6i), NE-SW Part of the study regions shows high iron content in the soils and remaining area having moderate values, based on Cu values, it has observed from the map (Fig .6j), lower values in the NE part and moderate in the center of the study region. Based on S values, it has observed from the map (Fig .6k), SE, and SW parts having low values and NW part having moderate and center of the study region showing high values.
3.4 Correlation between Physico-chemical parameters
To determine the relationship between the soil salinity and DN values for the total pixels of the different soils, an effort was completed to intercalate the soil salinity data. To interpolate the available soil salinity data, for the 11 soil samples have been observed with different soil parameter observations at the topsoil derived from soil mapping were digitized and then rasterized. The rasterized map was then interpolated using ArcGIS software which performs IDW based on the values of soil parameters. The correlation examination has been given in Table 3. “Based on the results attained from the soil analysis, an attempt was made to establish correlation for band 3 (green), band 5 (NIR), and band 7 (SWIR) of the LANDSAT 8 satellite image for the year 2019 with 11 soil parameters (pH, EC, SOM, N, P, K, Zn, Mn, Fe, Cu, and S), the DN of corresponding sites was statistically correlated with Physico-chemical parameters at the 0.5 significance level. There is no significant relation between visible band 3 and SWIR band 7 of the satellite image, hence those values are not mentioned here. The DN values used in the correlation matrix were the same as the values of band 5 (NIR)”. Nonetheless unveiled insignificant correlation with Physico-chemical parameters linked to the rest of the bands of the pixels for the 2019 image (Ahmad, N., & Pandey, P. 2018).
Rendering to our outcomes, the correlation between coefficient amongst the soil salinity and related DN values using LANDSAT data was supportive in calculating the significant relation between satellite data and soil salinity. The salt-affected soils in semi-arid regions show a high reflectance, especially when a salt crust (whitish color) is formed. For the assessment of LD and desertification, the correlation between DN’s and Soil parameters by concocting soil maps from remotely sensed data.