FORECASTING LAND SURFACE TEMPERATURE USING ARTIFICIAL NEURAL NETWORK

Visualizing the pattern of urbanization and correlating it with Land Surface Temperature (LST) serves as vital information for understanding the phenomena of urban heat island and other heat-related issues. LST is an important variable to define microclimate, ecology, bio-geo-chemical and biodiversity of the region. The foremost objective of this study is to forecast LST using Artificial Neural Network (ANN) and geospatial technology as a tool. Temporal land use and LST for 1991, 2000, 2009 and 2017 along with the elevation details were used to define the pattern followed by deriving relationship to forecast LST. The results obtained signifies a relationship between rise in concrete area and reduction in open and vegetated spaces with rising surface temperatures. The forecasting equation developed from the model shows good accuracy for prediction. Outcomes of the study demonstrated the capability and proficiency of ANN models to forecast surface temperature considering various parameters for the complex and dynamic physical environment.