Assessment of LULC dynamics and its association with LST distribution and NDVI Using Geospatial approaches in Karnataka state, India

Abstract

Monitoring land use and land cover (LULC), which changes at regional levels, is required for many kinds of applications, including monitoring of landslides, drought, flood, land erosion, agricultural planning for land, and climate change studies. The MODIS, Landsat-8, and Sentinel 2A satellite data are used for this investigation to extract the LULC, LST (Land Surface Temperature), and NDVI (Normalized Difference Vegetation Index) from 2015 to 2022. The LULC is performed using an advanced Google Earth Engine (GEE) tool that extracts LULC classes with specific training points of LULC classes. The two main machine learning approaches used for generating the LULC maps are Random Forest (RF) and Support Vector Machine (SVM). The Agricultural land (67.70 %), fallow land (1.76 %), forest land (20.04 %), built-up land (2.58 %), water bodies (5.95 %), waste land (6.78 %), and others (1.17 %) make up the majority of the study area in this class. In 2022, the largest occupied agricultural land area will be approximately 128615.8 km² compared to other classes. The NDVI and LST are the key indices to evaluate the vegetation and temperature (both seasonal and annual) of the region; these parameters are connected with LULC to study regional-level changes. The LST highest is in highest in 2021 is about 335.36 K (62.24 °C), and the lowest recorded in 2019 is 291.27 K (18.12 °C). The NDVI Value is higher in the South West monsoon season, especially in the Western Ghats, and the lowest record is in the north east part of Karnataka. This study is highly useful for the management of semi-arid regions, LULC categorization, forest ecosystem, environmental preservation, sustainable agriculture, controlled development, water shortage, and water management programs in the state.