| Summary: | The impacts of climate change are of increasing interest to water resources managers since the changes in water resources affect many fields such as agriculture, ecosystem, water quality and quantity etc. The typical framework of climate change impact studies on water resources is based on the utilisation of climate models and hydrological models. This thesis aims to explore and fill in the knowledge gaps in this framework. The Thorverton catchment (606 sq. km) in Southwest England is chosen as the case study catchment. The studies are composed of two main parts: bias correction of regional climate model (RCM) and hydrological modelling under climate change. Firstly, I propose four new bias correction schemes. The first two studies are related to grouping criteria and the other two are related to the data uncertainty: 1) an improved bias correction scheme based on comparative precipitation characteristics; 2) the Optimal number of bias correction groups; 3) bias correction methods for RCM simulations considering the distributional parametric uncertainty underlying the observations; and 4) precipitation ensembles conforming to natural variations derived from Regional Climate Model using a new bias correction scheme. Secondly, a new parameterisation scheme for the nonstationary hydrological system is explored. The calibration is based on changing the model parameter with time by adapting the parameter. Only one parameter is selected for optimisation while the other parameters are fixed. It is concluded that the performance of the proposed method is better than the conventional method whose parameters are stationary. In addition, calibration of non-continuous time series has been explored. Currently there is no consensus method on how to calibrate this non-continuous time period. I have explored two sub-annual calibration schemes (serial and parallel) and recommended that the right choice is dependent on the purpose (e.g. interested in soil moisture or flow) of the model usage.
|
|---|
