Projected near-term changes in monsoon precipitation over Peninsular Malaysia in the HighResMIP multi-model ensembles

Abstract

Changes in the monsoon season rainfall over Peninsular Malaysia by the mid-21st century are examined using multi-model ensemble data from the CMIP6 HighResMIP experiments. We examine simulations of the present and future climate simulations run under a high emission scenario of greenhouse gases from the Shared Socioeconomic Pathways (SSP5-8.5). The combined effects of horizontal and vertical resolutions on the projected changes in monsoon rainfall and associated environmental fields are investigated by comparing the ensemble mean of the projected changes utilizing appropriate multi-model groupings. The results indicate a projected decrease (by up to 11% near Mersing of eastern Johor, for the period 2031–2050 relative to 1981–2000) in monsoon precipitation along the southeastern coast of Peninsular Malaysia during the northeast monsoon season associated with the projected weakening of the monsoon flow during boreal winter. For the northwestern regions (e.g. Perak) often affected by severe floods, a significant increase in precipitation (by up to 33%) is projected during the southwest monsoon season, partly driven by the projected strengthening of the cross-equatorial flow and the weakened low-level anti-cyclonic shear of winds in boreal summer. However, the magnitudes and signal-to-noise ratios of the projected changes vary considerably with respect to different horizontal and vertical resolutions. Firstly, models with relatively high horizontal and vertical resolutions project a more significant decrease in precipitation during the northeast monsoon season. Secondly, for the southwest monsoon season, models with relatively high horizontal resolutions project larger magnitudes of increases in precipitation over the northern region, while smaller increases are found in simulations with relatively high vertical resolutions. Generally, reduced ensemble spread and increased signal-to-noise ratios are found in simulations at higher horizontal and vertical resolutions, suggesting increased confidence in model projections with increased model resolution.