CMIP5 wind speed comparison between satellite altimeter and reanalysis products for the Bay of Bengal

A proper evaluation and performance assessment of climate model projections have received considerable attention during the recent past amongst the scientific community. Quality of wind datasets used for analysis is of paramount importance to meteorologists, oceanographers, and climatologist as an essential pre-requisite for modelling needs. This study examined the measured wind speeds obtained from satellite altimetry available from IFREMER/CERSAT, along with two atmospheric reanalysis products ECMWF ERA-Interim and NCEP-CFSR. The reanalysis products and altimeter data were compared with wind speed simulated from 33 different models under WCRP-CMIP5 project for the Bay of Bengal (BoB) region. Study investigated both historical and projections of CMIP5 data providing an opportunity to inter-compare the wind speeds resulting from various emission scenarios with Representative Concentration Pathways (RCPs) 2.6, 4.5, 6.0, and 8.5, respectively. The objective is to establish and find out a suitable emission scenario applicable to the BoB region. Temporal and spatial analyses of CMIP5 data infer variability in terms of correlation, bias, and root mean square error. For the historical runs (1991–2005) based on analysis of 29 CMIP5 models, it could be ascertained that the correlation coefficient in wind speed varied between 0.6 and 0.9 and with a bias ranging from − 1.6 to 4 ms−1. Similar analysis of the CMIP5 projections was carried out with 11 models for RCP 2.6, 29 models for RCP 4.5, 10 models for RCP 6.0, and 28 models for RCP 8.5. Basin-scale mean using altimeter and re-analysis products indicates that RCPs 2.6 and 6.0 showed less correlation with a higher bias for the study region. Analysis of historical model runs signifies that HadGEM2-ES, HadGEM2-AO, HadGEM2-CC, MIROC5, GISS-E2R, and CNRM-CM5 are the best performing models for the study domain. Findings from the study indicate that RCP 4.5 wind speed stands better for the Bay of Bengal region. In a broader perspective, due to various uncertainties involved in climate model outputs, it is imperative to perform a comprehensive analysis amongst multiple data sources to establish and identify the best quality data for scientific needs.