Evaluation of COWCLIP2.0 Ocean wave extreme indices over the Indian Ocean

This study evaluates the performance of 39 CMIP5 models participating in the Coordinated Ocean Wave Climate Project phase 2 (COWCLIP2.0) for simulating extreme significant wave height (SWH) indices in the Indian Ocean (IO) for the 26-year period from 1979 to 2005, using the ERA5 wave reanalysis as observation proxy. The multiple skill metrics of bias, root mean square error (RMSE), relative error (RE), interannual variability skill-score (IVS), comprehensive rating index (CRI), and total ranking (TR) are utilized to evaluate the CMIP5 models consisting of four clusters (ECCC(s), CSIRO, ECCC(d), and JRC) over the Northern IO (NIO), SouthernTropical IO (STIO), and Southern IO (SIO) sub-domains. The three extreme SWH indices are considered: rough wave days (HsRo), high wave days (HsHi), and top decile wave spell duration indicator (HHsDI). Climatology evaluation results indicate that the ECCC(s) cluster models and MME exhibit better agreements with the ERA5 reanalysis data (with smaller biases, RMSEs, and REs) than the other clusters over all sub-domains for HsRo and HsHi indices. Whereas most models display reasonable skills at simulating interannual variability of HsRo, HsHi is poorly captured by all clusters over the NIO and STIO, with a large inter-model spread in IVS values. HHsDI is found to be simulated well by all clusters regarding the climatology pattern and interannual variability, reflecting the characteristics of percentile-based indices. Integrated assessment based on CRI and TR analysis confirms the overall superiority of ECCC(s) cluster models in simulating mean and interannual variability of extreme SWH indices over all IO subdomains.