APPLICATION OF ARTIFICIAL NEURAL NETWORK IN FORECASTING LAND SURFACE TEMPERATURE IN KARU LGA, NIGERIA

Abstract

Land Surface Temperature (LST) is a significant factor in the relationship between regional microclimatic changes and other environmental factors. This study aims to look at the LST's intensity during three time periods (2002, 2012, and 2022). Second, to predict the LST in Karu, Nigeria, for the years 2032 and 2042. The analysis used geospatial technology and an artificial neural network (ANN). On all maps, LST intensity was depicted. These classifications were based on a +3°C temperature rise within a range of 25°C to > 37°C. The results obtained signifies that classified LST for 2002 revealed a larger portion of area had surface temperature within the ranges of 28° - 31°C and 31° - 34°C covering 52.77% and 36.75% of the total study area. From 2012 and 2022; surface temperature within the range of 28° - 34°C found to dominate the spatial extent of the study area. The rise in urban areas and decline in vegetation cover may be related to this. As a result, an incremental tendency towards high LST being dominant throughout the research area's geographical extent was realised. By 2032 and 2042, 46% of Karu is expected to see temperature increases more than 38°C, following the LST trend. The study's findings confirm that ANN models are effective and capable offorecasting LST while taking into account a variety of characteristics for dynamic and complex real-world datasets. To reduce the occurrence of Urban Heat Islands (UH1) in the area, the research suggested sustainable urban development and increasing plant cover.