REGIONALIZATION OF BIAS CORRECTION METHODS FOR PRECIPITATION DATASET OVER INDIA

Abstract

Global climate models are fundamental research tools in climate change studies. While these comprehensive climate models address large-scale atmospheric features of climate effectively, the spatial resolution of these climate models on a global scale ranges from 150 to 300 km, which is very coarse for any region-specific studies. Though higher resolution global climate models can have extensive simulations for taking local climate aspects in their outputs, the high computational cost makes it unfeasible. Thus, regional climate models are used to dynamically downscale the Global climate model outputs to finer spatial resolution for streamflow simulation, climate change impact assessment, or future hydrological modeling. However, when comparing with the observed values, these downscaled climate model outputs obtained from RCMs are known to produce errors or biases in modelled climate outputs. Therefore, bias correction is needed to reduce systematic/ random biases for realistically representing climate data. For Indian conditions, even though a large part of India lies north of the tropic of cancer, the shutting effect of the Himalayas and the presence of the Indian ocean make Indian climate conditions complex. However, there are very few guidelines or studies that suggest the best bias correction method applicable to the region of India considering different climatic zones. Therefore, in this thesis work, bias correction factors were generated for RCM data from 1980-2005, and the results were validated using the observed data of 2006-2010 with a spatial resolution of 0.25⁰ x 0.25⁰ for CMIP5. Same methods were applied on CMIP6 GCM data from 1980 to 2014 and validated for the year 2015 - 2019 with 1⁰ x 1⁰ grid size over entire India for different scenarios. After segregating these grids into different climate zones, results obtained from validation suggested that parametric quantile mapping and distribution-based mapping method of bias correction are better performing than other methods in many climate zones.