Assessment of cadastral level climate extremes over tamilnadu india using deep learning technique

Abstract

Climate change is one of the significant issues faced globally in this century. It is unequivocal and unprecedented that the increase of greenhouse gases in the atmosphere and anthropogenic activities are the main drivers of global warming and the changing climate. The impact of these changes is manifested by the variation in duration, intensity and magnitude of extreme events. Due to the severe effects of extremities, the assessment of extremes has received greater attention in the scientific communities. Tamil Nadu, an important state in India for agriculture, is recurrently exposed to floods, cyclones, and droughts, which have devastating effects on humans, agriculture and the economy. This persuades the study to focus on the future climate and its impacts on the Tamil Nadu state. The future climate of the state is evaluated from the daily maximum temperature (Tmax), minimum temperature (Tmin) and rainfall simulations developed using Hadley Center s regional climate model (RCM) termed PRECIS (Providing REgional Climates for Impacts Studies) and Regional Climate Model version 4.4 (RegCM4.4) constituted by Abdus Salam International Centre for Theoretical Physics (ICTP). This study utilizes the lateral boundary condition generated from the Global climate models (GCMs) of the Hadley Centre Global Environmental Model (HadGEM2-ES) for PRECIS and GFDL, MIROC5, and CCSM4 for RegCM4.4 to downscale to the high resolution (horizontal resolution of 25 km) RCMs for the period of 1971-2094 under RCP 4.5 scenario. The model simulated variables such as daily Tmax, Tmin and rainfall are validated with the India Meteorological Department (IMD) dataset for the baseline period of 1971-2000 to assess the performance of the models. newline