Spatial-temporal Variation And Distribution Characteristics Of Droughts

In recent years due to the natural conditions, climate changes and human factors,Tianjin city occurs water resource crisises and drought disasters more frequentlywhich affects the economical and social development of Tianjin. Therefore, on thebasis of the monthly precipitation data, this article studies on the spatial and temporaldistribution of the drought in Tianjin and the drought variable combination probabilityand the return period is calculated and analyzed in order to provide the basis ofplanning the water resources and formulating the drought resistant measures. Themain research contents and results are as followed:(1) Based on the nearly50years’monthly precipitation data, the monthly andseasonal scale standardized precipitation index is calculated to conclude the temporaldistribution. The results show that drought frequency shows ascendant trend, droughtlevel is gradually too slightly and severe drought and the seasonal drought is moreobviously. Using Kringing spatial difference theory, the countor map of SPI wasdrawed. Tianjin city was divided to arid district by K-means clustering method. Theresults show that Tianjin city can be divided into6districts, which provided the basisfor formulating drought resistant plan.(2) Based on the theory of run, drought duration、drought severity and severitypeak of Tianjin were ectracted. Through the test and optimization of the edgedistribution function model, the drought durations are subjected to exponentialdistribution; drought severities are subjected to Gamma distribution and intenditypeaks are subjected to generalized Pareto distribution. Through the analysis of thecorrelation, the correlation between drought duration and drought severity、droughtseverity and severity peak of Tianjin is high, while the correlation between droughtduration and severity peak is low.(3) Using the maximum likelihood method to estimate the unknown parametersof copula coupling funtion, selecting the K-S statistic、OLS and AIC to test and dogoodness-of-fit evalution can prefer a copula coupling function which fitting thedrought variable best. The two-dimensional and three-dimensional drought variablemodel were established by the optimized copula function to calculate the jointprobability and joint return period and draw the corresponding graphics for inquringthe joint probability and joint return period under different combinations.(4) For two dimensional distributions, co-occurrence return period is muchgreater than the joint return period under the same condition. For three dimensionaldistribution, the co-occurrence return period is much greater than the joint returnperiod when the drought variables take the same value under the condition of given one drought variable; the joint return period of two drought variables under thecondition that other drought variable is greater than a certain value is longer than thatunder the condition of the other drought variable being smaller than the same certainvalue under the condition of given two variables.