[A-AS04] Extreme Events and Mesoscale Weather: Observations and Modeling

[AAS04-P06] On the Way to Assess and Identify: Do CMIP6 Climate Models Effectively Simulate Regional Droughts?

*Amit Dubey¹、Deepak Swami¹、Vivek Gupta¹、Nitin Joshi² (1.School of Civil and Environmental Engineering, Indian Institute of Technology Mandi, India 、2.Department of Civil Engineering, Indian Institute of Technology Jammu, India)

キーワード：Drought, CMIP6 models, Performance evaluation, Symmetrical uncertainty, SPEI, Indus river basin

Climate change is one of the giant global issue which adversely affects the water resources, agricultural productivity, human health, ecosystems and economy of the country. Above all, for a specific region, replication of the actual or realistic representation of climate characteristics (complex in nature) remains a great challenge. These in-situ climatic replications for historical variations and future drought prediction, heavily rely on climate models. The performance of General Circulation models (GCMs) released recently by World Climate Research Program (WCRP) under Coupled Model Inter-comparison Project phase 6 (CMIP6), in simulating droughts poses a challenge in comprehension due to the association of large uncertainties. In order to enhance the accuracy of future drought projections, best performing models are to be identified. Hence, evaluation of CMIP6 models becomes necessary to understand the inherent biases of the model simulation, before exploring the behavior of future droughts. Therefore, the present study aims to assess the performance of different CMIP6 models in simulating historical drought against the observations and identify the best performing models, over the Indian extent of Indus river basin during the period 1979-2014. For this purpose, gridded precipitation, maximum and minimum temperature data of 17 CMIP6 climate models is utilized to understand the drought simulations. Their performance has been carried out against the observed Indian Monsoon Data Assimilation and Analysis (IMDAA) reanalysis dataset (0.12 degree spatial resolution) obtained from National Centre of medium range weather forecasting, under the Ministry of Earth Sciences, Government of India. Quantification of drought is done on the basis of Standardized Precipitation Evapotranspiration Index (SPEI) at a time scale of 3 months to capture the meteorological droughts. An entropy-based robust feature selection approach known as symmetrical uncertainty (SU) is employed in order to rank the GCMs. Results revealed that models performance show high heterogeneity in simulating droughts along with the meteorological parameters viz. precipitation and temperature. KACE1-0-G, E3SM and INM-CM5-0 are found to be the best performing models for historical drought simulation based on SPEI, while ensemble mean of all the considered models along with E3SM and NESM illustrated best replication both for the precipitation and temperature respectively. Thus, it is evident from the study that different models should be taken into consideration for the future predictions of drought, because one which performs best in simulating parameters does not guarantee the same performance in drought simulation. The best suited models identified from this historical simulation experiments, ultimately will lead to the better prediction of future drought events under different climate change scenarios over the Indus river basin.