Exploring Factors Affecting Land Surface Temperature Using Machine Learning Technique: A Case Study on Meghalayan Terrain

Earth's surface temperature continues to rise steadily over time. A Long-term assessment of Earth's Land Surface Temperature (LST) trend and changes in land use/land cover (LULC) can help understand its cause and effect on the environment. This study aims to use LST trend and LULC change information to analyze its impact on the state of Meghalaya. As an indicator, the trend in LST is used to examine the effects of natural and human-induced activities, supported by the decadal land use change assessment. The study utilizes a Mann-Kendall test to identify the trend in LST from the last 22 years (2000-2021). Similarly, a machine learning technique (Random Forest) is used to analyze the dynamics of decadal LULC change. Inter class differences in conjunction with LST trends are then cross-validated to draw any conclusion. The test suggests an increasing trend in LST for the Meghalayan region. Corresponding analysis of LULC change suggests that the conversion of vegetation land to buildup areas rose from 28% in 2000-2010 to 45% in 2010-2021. Similarly, 20% of forest land was converted to buildup areas in the first half, with a slight decrease to 14% in the second half. Furthermore, the importance of temperature, rainfall, and DEM in modelling LST variations was also outlined as a part of the parameter importance analysis performed in this study. By utilizing these insights and methodology, effective measures can be developed to counter the harmful effects of deforestation and urbanization, thus preserving the ecological balance of our ecosystems for future generations.