The Simulation and Prediction of Land Surface Temperature Based on SCP and CA-ANN Models Using Remote Sensing Data: A Case Study of Lahore

Abstract

Over the last two decades, urban growth has become a major issue in Lahore, accelerating land surface temperature (LST) rise. The present study focused on estimating the current situation and simulating the future LST patterns in Lahore using remote sensing data and machine learning models. The semi-automated classification model was applied for the estimation of LST from 2000 to 2020. Then, the cellular automata-artificial neural networks (CA-ANN) module was implemented to predict future LST patterns for 2030 and 2040, respectively. Our research findings revealed that an average of 2.8 °C of land surface temperature has increased, with a mean LST value from 37.25 °C to 40.10 °C in Lahore during the last two decades from 2000 to 2020. Moreover, keeping CA-ANN simulations for land surface temperature, an increase of 2.2 °C is projected through 2040, and mean LST values will be increased from 40.1 °C to 42.31 °C by 2040. The CA-ANN model was validated for future LST simulation with an overall Kappa value of 0.82 and 86.2% of correctness for the years 2030 and 2040 using modules for land-use change evaluation. The study also indicates that land surface temperature is an important factor in environmental changes. Therefore, it is suggested that future urban planning should focus on urban rooftop plantations and vegetation conservation to minimize land surface temperature increases in Lahore.