Probability And Statistical Analysis Of Flood Extreme Events Under Climate Change In The Huaihe River Basin, China

Climate change is altering the global hydrologic cycle and the spatio-temporal evolution of water resources, which affects the frequency and intensity of hydrological extreme events. In recent decades, flood disasters occurred frequently in China, bringing huge losses to the people's life and property, which is a serious threat to regional water resources security. Thus, the research on statistical characteristics and spatio-temporal variations of flood extreme events under climate change has important scientific value and application meaning for the regional disaster prevention and mitigation, adaptive management of water resources and sustainable development of society and economy. The dissertation takes a typical basin the Huaihe River Basin (HRB) as the study area, revealed the spatio-temporal evolution and cycle characteristics of flood extreme events, analyzed the univariate and multivariate probability statistical characteristics of flood extreme events under climate change, and further studied the influence on frequency and intensity of flood extreme event by climate change in the Huaihe River Basin via the statistical model and hydrological model. The main work and research results of the dissertation are as follows:(1) Spatio-temporal variations of the flood extreme events in the HRBThe Mann-Kendall test, moving T-test method and wavelet transform analysis method are applied to analyse the spatial and temporal variations of the precipitation, extreme precipitation and flood extreme events were analyzed, as well as the flooding events during the recent500years in the HRB. Results show that:During1956-2010, the annual precipitation in the HRB shows an increase trend, which is not obvious. Extreme precipitation events mainly occur during June to September, accounting for more than75%of the total precipitation in all the year round. The extreme precipitation indices such as the annual extreme precipitation volume, days and intensity in the HRB show a trend of increase. Most of the maximum flood peak in the HRB appears in the1960s-1970s, and concentrates in the flooding season. The number of flooding events with return period over50years is the most in the1960s, then decreases. The flooding events with return period between20to50years occurs most frequently in the1980s, then decreases quickly in the1990s, while it shows an increase trend in the21st century. During1470-2010, except Xinyang, Bengbu, Fuyang, Xuzhou and Linyi are flooding overall. Meanwhile, the flooding events at these four stations have significant periodicity of40and90years.(2) At-site and regional frequency analysis of the flood extreme events in the HRB Based on the linear moment theory, the annual maximum (AM) and peaks over threshold (POT) series of flood extreme events in the HRB are analyzed via at-site and regional frequency methods, and the accuracy and uncertainty of estimated quantiles through these two methods are also studied and compared. Results show that: The flood peak is highest in the main stream of the HRB, which increase gradually from the upstream and that on the right bank of the Huaihe River i.e. Huainan mountainous area is lager than that on the left bank. The flood volume and duration become larger gradually from the upstream, which have good positive correlation with the subbasin drainage area. So both the results from at-site and regional frequency analysis are consistent of the geographical and flood characteristics of the HRB. For regional frequency analysis, the biases of the estimated quantiles become larger with the increase of the return period. When the return period is over100years, the estimated quantiles results become unreliable. Compared with the at-site frequency analysis, the estimated quantiles results from regional frequency analysis is more accurate with less uncertainty, especially for the high quentiles estimation. The POT series can describe the flood extreme events in the HRB better than the AM series.(3) Multivariable frequency analysis of flood extreme events in the HRBTwo dimensions and three dimensions multivariable statistical models are established to describe the flood extreme events in the HRB via Copulas. Furthermore, the meteorological elements are taken into consideration, binary statistical model are build between previous cumulative precipitation and AM peak to study the influence on the flood extreme events by the change of meteorological factors. Results show that:Compared with maximum pseudolikelihood technique involving the ranks of the data (MPL) and inference function for marginal method (IFM), moment-like method of inversion of association measures (MOM) method is the best for estimating the parameters of Copulas, which is robust. The selected optimal Copulas can describe not only the dependence of flood extreme events variables but also the tail dependence. With the increase of the previous cumulative precipitation, the possibility of small flooding events decreases gradually, but the possibility of extreme flooding events increases gradually, and the possibility of medium and big flooding events increases first and then decreases gradually.(4) Flood extreme events in the HRB under climate changeThe nonstationary frequency analysis of the flood extreme events has been made through Generalized Additive Models for Location Scale and Shape (GAMLSS) statistical model, and Time Variant Gain Model (TVGM) and future climate patterns of IPCC AR4 are used to estimate the flood extreme events of the HRB under the future climate scenarios. Moreover, the spatial and temporal variations of the flood extreme events under the future climate scenarios are analyzed and the influence of climate change on the frequency and intensity of the flood extreme events are studied. Results show that: GAMLSS model can describe the nonstationary series of flood extreme events of the HRB better than the stationary distributions. TVGM has good effect and applicability for simulating the runoff in the HRB. In the three climate scenarios A1B, B1and A2, the annual precipitation volume show an increase trend in the basin. The AM peaks at most stations show a decrease trend during1956-2099, and the oppsite situation during2020-2099. In the future scenarios, AM flood peak become smaller than the history values. Under the different return periods, at most stations AM flood peak under the A2scenario is the largest, which is smallest under B1scenario, and in the middle under A1B scenario. Compared with historical flood extreme events, the occurrence of big and medium flood reduce, especially the medium flood event decreases significantly. However, the occurrence of small flood events increases. Relative to history condition, under the future climate scenarios AM flood peak intensity decreases.