STATISTICAL MODELLING AND ANALYSIS FOR REGIONAL CLIMATE CHANGE

General circulation models (GCMs) are often used to inform climate change information, however, due to their coarse spatial resolutions, downscaling approaches are often relied to transit the coarser-scale GCM outputs to higher resolutions. Multivariate multisite weather generators (MMWGs) are appealing tools, as they allow for simulations of multiple realizations of climate change scenarios consistent with local-scale weather characteristics as well as large-scale climate change signal informed by GCMs. This thesis developed a new MMWG which integrates a single WG with a post-processing technique to rebuild the observed inter-site, inter-variable correlations, and temporal structures in the simulations. The proposed MMWG was then adapted and extended to a multivariate multisite statistical downscaling (MMSD) approach for regional climate change study. Along with these statistical modeling approaches, observational investigation of the climate change signal in the gauge-based precipitation observations was also carried out. The proposed approaches could add value to modeling of regional climate change.

Author dc:creator

• LI XIN

• Weather Generator, Statistical Downscaling, Rainfall Disaggregation, Wet and Dry Spells, Rainfall Extremes, Singapore