Trend analysis of rainfall projections of Teesta river basin, Sikkim using non-parametric tests and ensemble empirical mode decomposition

This paper presents the spatio-temporal trends and dominant change points in the Regional Climate Model system (RegCM) simulations at 0.43 degrees x 0.43 degrees resolution over Teesta river basin located in Sikkim. In this study, first the bias correction is done following the Local Intensity Scaling (LOCI) method by comparing the interpolated model values with the observed values of five stations for the historical period of 1983-2005. The estimated correction is applied to the interpolated future RegCM data for the rainfall projections of 2021-2050 period under two Representative Concentration Pathway (RCP) scenarios (RCP4.5 and RCP8.5). The historical rainfall and rainfall projections of the near future period of 2021-2050 are subjected to trend analysis using Mann-Kendall method and its sequential version (SQMK), Sen's slope estimator, linear trend fitting and Ensemble Empirical Mode Decomposition methods. The preliminary estimates using non-parametric tests showed a likely reversal in nature of trend in Gangtok and Kalimpong stations. The results of SQMK test detected possible trend turning points in 2030s except in future rainfall of Darjeeling station. The results of non-linear trend analysis showed that the future rainfall of Jalpaiguri shows an increasing trend while that of Gangtok shows decreasing trend irrespective of the two candidate RCP scenarios. The non-linear trend is different from linear trend in rainfall of Lachung and Kalimpong stations under both scenarios and that of Jaipalguri station under RCP8.5 scenario. The extraction of true shape of inherent non linear trend performed in this study may help for improved predictability and better management of water resources of Teesta river basin.