Hydrological Drought Indices And Frequency Analysis Methods And Their Applications

Drought is a frequently occurred natural hazard,which is always recognized as wide-spread and long-lasting across the world,and limits the development of society,economics and human civilization.Deeply understanding the triggering mechanisms and evolution rules of drought,as well as precisely depicting its quantitative characteristics and multi-dimension properties are of great importance to improve our capability of hazard warning,risk management and regional water allocation.In this thesis,the drought definition and drought indices,as well as drought characteristics analysis methods and drought assessment methods under changing environment were comprehensively reviewed.The Hanjiang River Basin,which is the experimental area of the most stringent water resources management system in China,was chosen as case study.The thesis mainly focuses on hydrological drought indices and frequency analysis methods.The main conclusions and innovations were summarized as follows:(1)The Principle of Maximum Entropy(POME)method was used to model the probability distribution functions of aggregated monthly streamflow with varying time scales for 30 sub-basins of the Hanjiang River Basin.The first three original momonts of the cumulative monthly streamflow data were chosen as the constraint functions for maximizing the entropy.The Streamflow Drought Index(SDI)was computed based on monthly streamflow records assuming sampled from several theoretical probability distributions,e.g.POME,Normal,Log-normal,Weibull and Pearson Type ?.By maximum usage of the information from observed records while avoiding mistakenly envolving redundant information,the POME-based probability distribution functions show satisfying applicability.It is found that the probability distributions of cumulative monthly streamflow will change towards normalization with the increasement of the time scales.(2)A modified parsimonious two-parameter monthly water balance model was proposed to simulate the monthly hydrological processes of 30 sub-basins of the Hanjiang River Basin.The probability distribution functions of monthly runoff were derived based on the marginal distribution functions of climatic inputs(i.e.monthly precipitation and potential evapotranspiration)and antecedent generalized soil moisture using the integration by substitution and copula method.Then the Standardized Runoff Index(SRI)was calculated.With the acceptable performance of the hydrological model,the derived distribution is competable to the conventional curve-fitting mothod.By investigating the correlation between four major drought characteristics(i.e.drought duration,drought severity,drought intensity and interarrival time)and four dimensionless parameters representing the climatic and underlying properties of the basin(i.e.Aridity Index,Storage Index,Time-scale Transformation Index for Evapotranspiration and Runoff Coefficient),conclusion can be drawn that the formation and development of hydrological drought in Hanjiang River Basin is mainly controlled by the watershed storage factors.However,the influence of climatic factors can not be neglected.(3)A simulated-based water resources allocation model considering the water supply and reservoir operation were proposed to reconstruct the monthly runoff of two stations along the mainstream of the Hanjiang River.With the satisfying accuracy of the water resources allocation model,a time-varying moment method was used to perform nonstationary frequency analysis for monthly runoff records taking the agricultural water comsumption and a Reservoir Index as covariates.A Nonstationary Standardized Runoff Index(NSRI)was thus derived based on the nonstationary monthly runoff quantiles.It is concluded that the time-varying distribution is necessary for calculating the standardized drought index in the case of nonstationary monthly runoff.It is also found that the NSRI is more sensitive to the sereve and extreme drought events.(4)The propagation direction and magnitude among different drought types,i.e.meteorological,agricultural and hydrological drought,represented by different standardized drought inidices,were assessed by an information-based transfer entropy method.It can be observed that the meteorological droughts are originated from the single or combined anomaly of precipitation decifit and temperature rising.Meteorological drought can develop into agricultural droughts represented by decreasing soil moisture,and hydrological droughts represented by decreasing runoff,respectively.At the same time,the soil moisture storage of the basin can have a feedback to the meteorological droughts in turn throuth land-atmosphere coupling.However,the propagation properties between agricultural and hydrological droughts are far more complicated since they are always synchronizedly related with each other,as well as influenced by the basin storage capacity.(5)The theoretical probability distribution functions of hydrological drought duration and severity,assuming geometric and Gamma distribution,respectively,were analytically derived based on the expectation and variance of annual runoff with Log-normal,Gamma and Normal distribution.The return period of hydrological drought event larger or equal to a specific drought severity was derived under the concept of Average Elapsed Time.The derived return period overcomes the inadequency of conventional return periods calculated by the inferential approach when assessing the risk of a multi-year drought.The applicability of the derived distribution functions was validated by the Monte Carlo simulation test with acceptable accuracy,especially for hydrological events with relatively short return periods.The occurrence and magnitude of hydrological droughts with specific dimensionless severity are spatially heterogeneous,which is mostly determined by the coefficient of variation of the underlying annual runoff.The difference between the return periods of a same dimensionless severity is increased with decreasing trunctive threshold used to identify a drought event by the run theory.