Investigation of Urban Heat Islands and modeling of Land Surface Temperature over selected Indian cities using MODIS products

Investigation on Land Surface Temperature (LST) and Urban Heat Island (UHI) is essential for cities at various stages of urban development, which is often overlooked. This study examines the UHI effect at selected Tier 2 cities in India with varying size and developmental stages. MODIS products such as MOD11A1, MOD09A1 and MCD12Q1 for the duration 2006–2022 were used for computing various spectral indices along with corresponding Land Use Land Cover (LULC) and LST. Results indicated that the highest UHI effect (above 3 °C) was experienced in Gurgaon (heavily built-up, scarce vegetation within core urban area) during summer nights. Small, unplanned cities like Dahegam and Kalol in Gandhinagar district experienced higher LST and UHI effect than the nearby planned state capital, Gandhinagar. On the other hand, the small, refinery city of Bongaigaon indicated Urban Cool Island (UCI) effect even during summers, due to the presence of vegetation within its core urban area. The potential of two network based algorithms – Convolutional Neural Networks (CNN) and Recurral Neural Networks (RNN) in modelling LST using the MODIS derived spectral indices and LULC classes as inputs is compared. Both CNN/RNN modelled LST showed good agreement with observations with R2 values during model testing period as high as 0.940/0.997 for Gandhinagar District, 0.991/0.996 for Bongaigaon and 0.988/0.993 for Gurgaon Administrative Division. The analysis of LST variation and seasonal UHI effects may be a key to the development of sustainable communities not only in large cities but also the less investigated smaller cities with similar concern.