Climate Change Impacts On Water Resources Based On Statistical Downscaling

Due to the effects of natural and human factors, climate change is being intensified, which makes water resources and water cycle system suffering from more and more serious stress. It aggravates the tendency variation of river runoff and the frequency change of the extreme hydrological events, and then increases the uncertainty of regional and local hydrological risks and the difficulty of water conservancy project design. Therefore, it is urgency and necessary for the sustainable water resources management to make accurate long-term forecasts for runoff in watershed under the changing climate. In view of this, in this study, taken Xiangxi River watershed as an example, the changing trends of the observed hydrological and meteorological variables would be analyzed. On the basis of that, under the latest emissions scenarios, Representative Concentration Pathways (RCP), in the Coupled Model Intercomparison Project Phase 5 (CMIP5), the General Circulation Models (GCMs) outputs based on the bias-correction and spatial disaggregation (BCSD) method would be statistically downscaled and applied to generate the corresponding future climate scenarios for driving the Stepwise-clustered Hydrological Inference (SCHI) model to predict runoff under each scenario. By this way, this study aims to investigate the impacts of climate change on hydrology and water resources in the future, in order to provide valuable scientific basis for management decision-making. The main content and the research results of this study are as follows:(1) The changing trends of the observed hydrological and meteorological data within Xiangxi River watershed were analyzed. On interannual time scale, there were no obvious change trend in both precipitation and runoff series, while the series of temperature and evaporation both clearly present gradual increase tendencies. Besides, the effect of precipitation on runoff was the most significant among all the meteorological factors.(2) Based on statistical downscaling method, the Statistic downscaling model (SDSM) and Support Vector Machine (SVM) model were established, validated, and compared. For the two models, the simulations for temperature were generally better than those for precipitation, and SVM model has a more superior performance.(3) In the 4 kinds of RCP emissions scenarios, the future climate scenarios within Xiangxi River watershed were generated using the established SVM downscaling model and the results were analyzed. In the future, the peak of precipitation in all scenarios would often appear in June. Compared with the past, the precipitation would reduce significantly in July and August, while increase obviously from September to December. And the temperature would change a little in comparison to the past. Except for the slight decline in July and August, the temperature would generally rise in all the other months. Besides, in the summer, the precipitation and temperature are both lowest in RCP8.5 scenario than those in the other three scenarios, which is oppsite in the other seasons. On the time step of decades, the change of precipitation under each scenario would become more and more intense with the growing radiation force of greenhouse gases, but the trends would be different from each other. The temperature under each scenario would be on the rise in general, and the uptrend would have roughly positive correlation with the value of radiation force.(4) For the runoff prediction, the SCHI model and Artificial Neural Network (ANN) model were established, validated, and compared. The SCHI model shows a more superior performance.(5) Based on the acquired future climate scenarios, the SCHI model was applied to runoff forecast and evaluation in the future. The flow peak would appear in May rather than in July as previously. At the same time, the runoff would generally increase in January and March, and from September to December, while decrease significantly in April, June and August. This is mainly due to both the change of precipitation and the variation of evaporation caused by the rising and falling in temperature. On decadal time scale, the long-term trend of surface runoff will change more and more drastic and complicated along with the increasing radiation force of greenhouse gases in the future.