Abstract:

As a complex natural phenomenon，drought is stochastic and has multiattributes.The stochastic theory and method is an effective tool to study droughts，and application of copulas remedies the deficiency of univariate analysis and overcomes the limitations of conventional bivariate models.In order to reduce the loss caused by drought，we studied the hydrological drought characteristics and drought frequency of the Nanpan River Basin to provide a basis for drought prediction.We used the monthly data of runoff collected from 3 hydrological stations of the Nanpan River Basin in 52 years (1961-2012)，and calculated the Standardized Streamflow Index (SSFI) as the indicator.The hydrological drought characteristics of the Nanpan River Basin were analyzed based on the frequency of drought duration，drought intensity and drought peak in the recent 52 years.Moreover,the hydrologic drought frequency was analyzed through statistics by taking the drought frequency，drought duration，drought intensity，drought peak in the 1992-1993 year and 2011-2012 year，and return period of drought in the time scale as the indicator.The results indicated that the drought characteristic values of 3 hydrological stations of the Nanpan River basin were different，all of the largest drought peaks were more than 2 and drought becamfe very serious.From 1961 to 2012，the maximum of drought characteristic values appeared in the 2011-2012，and it was one of the most serious drought.From a single site，the drought frequency，maximum of drought duration and drought intensity of ZhanYi hydrological station were greater than Gao Guma and Jiang Bianjie hydrological stations.The hydrologic drought of ZhanYi became more serious than other hydrological stations，so it needs more attentions.In addition，the drought peak in the Jiang Bianjie hydrological station was the biggest.Looking from the joint sites，the drought characteristic values of drought duration，drought intensity and drought peak in the above middle were greater than below middle，but the characteristic values from the three hydrological stations fell between upstream and downstream，so the characteristic values of upstream became bigger，and the hydrologic drought was very serious.By using the RMSE，AIC and BIC evaluation，the results show that，for the entire basin sites，Frank copula constructed the paircopula had a better fit，and suitable for drought frequency analysis in the Nanpan River Basin.The second best copulas were Clayton and Gumbel copulas，while the fitting effects had no obvious difference.Beside AMH copula，the three paircopulas were all may be the optimal copula function for one site.Threedimension joint distribution models were constructed based on the best copulas functions，and the associated probability and return period of drought variables were calculated for Nanpan River Basin.The results of multivariate joint return period show that，the joint nonexceedance return period was less than cooccurrence nonexceedance return period ,the return period of 2009-2012 year was during the period of 100-200 years，which became one of the most severe drought in 52 years (1961-2012)