New Bias Correction Methods for Simulating Precipitation and Runo

Abstract

Climate change is a huge environmental issue that our society currently faces. This thesis develops and tests two bias correction methods for regional climate simulations of precipitation and runoff . Biases in the soil water physics are corrected by including new physics in the soil moisture parameterisation and the regional model inputs are corrected statistically. Case studies are performed on the Olifants River basin in the Limpopo region of South Africa using the Weather Research and Forecasting (WRF) regional climate model. Accurate knowledge of water availability in this water-stressed region is of great importance for adaptation and future water policy development. The concept of tightly bound water, in which a reservoir of soil water is held stationary within small soil pores but is still available for evapotranspiration, is parameterised for the first time within the land surface scheme of a regional atmosphere-land surface model. Results of a WRF simulation forced by re- analysis show that the standard NOAH land surface scheme over-estimates mean annual runoff by 120% with respect to observations, despite rainfall and atmospheric conditions similar to observed. Use of the tightly bound water scheme within the NOAH model reduces this bias to 22%. Simulations with the WRF model forced with 1980s and 2040s CCSM3.0 general circulation model data show that the tightly bound water scheme significantly reduces runoff in different climates. The new scheme projects a 10% decrease in runoff by the 2040s compared to a 4% decrease projected by the standard model. A new quantile-mapping bias-correction of inputs to regional climate models is proposed. Linear correction and quantile-mapping methods are implemented to correct CCSM3.0 data using re-analysis. Simulation results show a significant difference between the correction methods. The results indicate that the quantile-mapping correction method could be developed to help produce more accurate regional climate predictions for impact studies.