Impact Of Climate Change On Hydrology In Qiantang River Basin Based On The SWAT Model

Because of global climate change, the mean surface temperature of China has increased obviously in recent100years, which may influence the precipitation and water cycle. The frequency of some extreme events has increased, too. Up to now, climate change has become a global environmental problem, which is attracting more and more attention. Investigating the impact of climate change on hydrology and water resources could help decision makers in hydraulic project planning, water management, disaster prevention and mitigation.There are large uncertainties involved in climate change impact analysis. This thesis employs the commonly used techniques in climate change impact analysis including general circulation models (GCMs), downscaling methods, and hydrological models, etc. Firstly, based on a weather generator and frequency analysis, the impact of climate change on extreme precipitation is investigated. Secondly, the SWAT model is set up in an upstream sub-basin and the river runoff is simulated in monthly scale to investigate the impact of climate change on water resources. Thirdly, the SWAT model is applied on daily scale, together with the frequency analysis method, the impact of climate change on extreme flows is investigated and the uncertainties from emission scenarios, extreme value models and SWAT model parameters are considered. Finally, two methods to improve extreme flow simulation in the SWAT model, namely post-processing technique and multi-objective calibration are proposed. The main work and conclusions are as follows:(1) Under scenarios A1B, A2and Bl, the outputs from HadCM3, CCSM3and ECHAM5models are downscaled through a weather generator LARS-WG. Daily precipitation in three future periods is simulated for frequency analysis. The results show that the design rainfall in all future periods increases at most stations under the three GCMs and emission scenarios. However, there are large uncertainties involved in the estimations of design rainfall at seven stations. At Hangzhou Station, a relative change of-16%to113%can be observed in the design rainfall of100-year return period.(2) Downscaled GCM ensemble data is put into the SWAT model to simulate future river runoff in Lanjiang River Basin. The results show that the relative change of annual precipitation in the basin is from-2.62%to2.74%. But the relative change of monthly precipitation is obvious. For example, the largest increase and decrease at Jinhua station under A2scenario are13%and18%, respectively. The relative change of annual potential evapotranspiration is from1.63%to13.67%, indicating an increase trend in the future period. The relative change of annual runoff is from-21.72%to1.28%, indicating a decrease in the future period and less water resource possibly available for the region in future.(3) The regional climate model PRECIS is applied to downscale the HadCM3outputs and future daily flows are simulated by the SWAT model in Lanjiang River Basin. Three extreme value models are used in frequency analysis to assess the impact of climate change on extreme flows and the uncertainties involved are investigated. The results show that design discharges of small return periods in the future period2011-2040under different emission scenarios are likely larger than those in the baseline period. For Jinhua and Yiwufotang stations, the design discharges of large return periods are larger than those in the baseline period, indicating a possible increase of floods in these areas. The uncertainty analysis results show that the SWAT model parameters are a very important uncertainty source, followed by emission scenarios and extreme value models. However, the extreme value models may contribute more uncertainty in the design discharges with the increasing return period.(4) The SWAT and GR4J models are used to simulate daily flows and the peaks over threshold (POT) method is employed to extract the extreme flows. The quantile mapping (QM) method and two generalized linear models are used to set up the post-processing model. Besides, a multi-objective calibration method is used to improve the extreme flow simulation in the GR4J model. The results show that both post-processing and multi-objective calibration methods can improve the extreme flow simulation to some extent. The QM method performs better than the other two generalized linear models in post-processing. Multi-objective calibration of GR4J model could improve the extreme flow simulation, but may weaken the performance in daily flow simulation. In regard of the performance of post-processing and multi-objective calibration, it is hard to judge which is better.