Impact assessment of land-use alteration on land surface temperature in Kabul using machine learning algorithm

Abstract

ABSTRACT This research evaluates the impact of LULC changes on LST of Kabul City, Afghanistan using Landsat data and Machine Learning Algorithm. The Cellular Automata Markov (CA-Markov) and Artificial Neural Network (ANN) models were used for future predictions. Results showed a significant increase in built-up areas, such as 8.54%. However, the vegetation and bare soil were reduced by approximately 6.97% and 2.18% between 1990 and 2020, respectively. The maximum annual mean LST was found in built-up areas, followed by bare soil and vegetation, while the mean annual LST increased by about 3.52°C. According to seasonal analysis, LST was reported higher during the summer, followed by autumn, spring, and winter. Future predictions showed that built-up areas, which were 14% in 2020, are likely to increase to 18% and 20% in 2030 and 2040, respectively. The regions with higher annual mean LST class (>30°C) are expected to grow by about 58% and 70% in 2030 and 2040, respectively. This research would improve the urban planning to avoid any possible effects of Urban Heat Islands (UHIs).