Statistical downscaling of CMIP5 outputs for projecting future changes in rainfall in the Onkaparinga catchment

Md Mamunur Rashid; Simon Beecham; Rezaul K Chowdhury

doi:10.1016/j.scitotenv.2015.05.024

Statistical downscaling of CMIP5 outputs for projecting future changes in rainfall in the Onkaparinga catchment

Sci Total Environ. 2015 Oct 15:530-531:171-182. doi: 10.1016/j.scitotenv.2015.05.024. Epub 2015 May 28.

Authors

Md Mamunur Rashid¹, Simon Beecham², Rezaul K Chowdhury³

Affiliations

¹ Centre for Water Management and Reuse, School of Natural and Built Environments, University of South Australia, Mawson Lakes, SA 5095, Australia. Electronic address: mdmamunur.rashid@mymail.unisa.edu.au.
² Centre for Water Management and Reuse, School of Natural and Built Environments, University of South Australia, Mawson Lakes, SA 5095, Australia. Electronic address: simon.beecham@unisa.edu.au.
³ Centre for Water Management and Reuse, School of Natural and Built Environments, University of South Australia, Mawson Lakes, SA 5095, Australia; Department of Civil and Environmental Engineering, United Arab Emirates University, Al Ain, PO Box 15551, United Arab Emirates. Electronic address: rezaulkabir@uaeu.ac.ae.

PMID: 26026419
DOI: 10.1016/j.scitotenv.2015.05.024

Abstract

A generalized linear model was fitted to stochastically downscaled multi-site daily rainfall projections from CMIP5 General Circulation Models (GCMs) for the Onkaparinga catchment in South Australia to assess future changes to hydrologically relevant metrics. For this purpose three GCMs, two multi-model ensembles (one by averaging the predictors of GCMs and the other by regressing the predictors of GCMs against reanalysis datasets) and two scenarios (RCP4.5 and RCP8.5) were considered. The downscaling model was able to reasonably reproduce the observed historical rainfall statistics when the model was driven by NCEP reanalysis datasets. Significant bias was observed in the rainfall when downscaled from historical outputs of GCMs. Bias was corrected using the Frequency Adapted Quantile Mapping technique. Future changes in rainfall were computed from the bias corrected downscaled rainfall forced by GCM outputs for the period 2041-2060 and these were then compared to the base period 1961-2000. The results show that annual and seasonal rainfalls are likely to significantly decrease for all models and scenarios in the future. The number of dry days and maximum consecutive dry days will increase whereas the number of wet days and maximum consecutive wet days will decrease. Future changes of daily rainfall occurrence sequences combined with a reduction in rainfall amounts will lead to a drier catchment, thereby reducing the runoff potential. Because this is a catchment that is a significant source of Adelaide's water supply, irrigation water and water for maintaining environmental flows, an effective climate change adaptation strategy is needed in order to face future potential water shortages.

Keywords: Bias correction; Climate change; Downscaling; General circulation model; Generalized linear model; Rainfall.