Projections of future extreme rainfall events using statistical downscaling in Malaysia

Abstract

Climate change is one of the greatest challenges for water resources management. Intensity and frequency of extreme rainfalls are increasing due to enhanced greenhouse gas effect caused by climate change. A lot of research has been done in developing innovative methods for assessing the impacts of climate change on rainfall extremes. Climate change strongly depends on General Circulation Model (GCM) outputs since they play a pivotal role in the understanding of climate change. However due to their coarse resolution, statistical downscaling is widely applied to match the scale between the GCM and the station scale. This research proposed to establish statistical downscaling model that was able to generate hourly rainfall data for future projection of hourly extreme rainfall in Peninsular Malaysia. An Advanced Weather Generator (AWE-GEN) built on stochastic downscaling principles was applied for simulating hourly rainfall data. The model construction involved 40 stations over Peninsular Malaysia with observations from 1975 to 2005. To account for uncertainties, an ensemble of multi-model namely GFDL-CM3, IS-CM5A-LR, MIROC5, MRI-CGCM3 and NorESM1-M were obtained from the dataset compiled in the WCRP’s, CMIP5. The projections of extreme precipitation were based on the RCP 6.0 scenario (2081-2100). To address the problem of unavailability of rainfall data at remote areas over Peninsular Malaysia, this research also examined the spatial variability of rainfall and temperature parameters using Locally Weighted Regression. Results of the AWE-GEN showed its capability to simulate rainfall for Peninsular Malaysia. Both hourly and 24 hour extreme rainfall showed an increase for future. Extremes of dry spell was projected to decrease in future whereas extremes of wet spell was expected to remain unchanged. Simulations of present climate using interpolated parameters showed promising results for the studied regions.