Inter-comparisons and applicability of CMIP5 GCMs, RCMs and statistically downscaled NEX-GDDP based precipitation in India

Vishal Singh; Sanjay Kumar Jain; Pushpendra Kumar Singh

doi:10.1016/j.scitotenv.2019.134163

Inter-comparisons and applicability of CMIP5 GCMs, RCMs and statistically downscaled NEX-GDDP based precipitation in India

Sci Total Environ. 2019 Dec 20:697:134163. doi: 10.1016/j.scitotenv.2019.134163. Epub 2019 Aug 30.

Authors

Vishal Singh¹, Sanjay Kumar Jain², Pushpendra Kumar Singh²

Affiliations

¹ National Institute of Hydrology, Roorkee 247667, Uttarakhand, India. Electronic address: vishal18.nihr@gov.in.
² National Institute of Hydrology, Roorkee 247667, Uttarakhand, India.

PMID: 32380622
DOI: 10.1016/j.scitotenv.2019.134163

Abstract

The applicability of GCMs (produced at 2° to 4°) and RCMs (produced at ~0.5°) vary and they might produce lots of ambiguity in their outcomes, because of their resolutions. This is true fact and already been reported in several studies. In this study, we have explored the precipitation variabilities in India involved in different resolution climate model based data-sets such as GCMs and RCMs under two extreme Representation Concentration Pathway (RCP) experiments (e.g. RCP4.5 and RCP8.5). Precipitation inter-comparisons have been done between different resolution datasets (e.g. GCMs, RCMs, NEX-GDDP and observed precipitation) to explore precipitation ambiguity (or variabilities) and their applicability in a long-term period (1951-2100) across the India. The observed gridded precipitation (1951-2005) and NEX-GDDP datasets (2006-2100) have been used as a reference data-sets to assess the accuracy of GCMs and RCMs. Variations in precipitation trends have been explored at each grid scale utilizing non-parametric Mann-Kendall test and statistical p-value test at 95% confidence interval. Inter-comparison analysis results showed that a significant diversity existed in the precipitation amounts among all climate model datasets, which have been non-uniformly distributed across the India. Results from model inter-comparisons, percentage of change analysis and Q-Q analysis performed between GCMs versus observed precipitation, RCM versus observed precipitation, GCMs versus RCMs and RCMs versus NEX-GDDP models showed a high variability existed in precipitation amount across the India during1951-2100. In opposite, at several locations a good association in precipitation between different resolution datasets was observed. GCMs based precipitation was underestimated and RCMs showed overestimation across the India. Overall, RCMs based rainfalls have found comparatively closer to observed and NEX-GDDP based rainfalls, yet RCMs have highly overestimated in several regions of India. Seasonal trends uncertainty estimation showed a better correlation in precipitation between NEX-GDDP and RCMs, especially during monsoon and pre-monsoon season.

Keywords: GCM; India; NASA NEX-GDDP; Precipitation; RCM; Variability.