| Climate change and human activities can intensify the uncertainty of water resources supply that based on melt ice and snow in the arid region of northwest China,which results in the increase of frequency and intensity of snowmelt flood disaster,the change of spatial and temporal distribution of snowmelt flood runoff,and the inconsistency of measured hydrological time series.If the traditional method of flood frequency analysis were applied to hydrologic design of hydraulic engineering,the flood control risk of hydraulic engineering and the reliability of water resources utilization derived from design flood calculation would be severely undervalued or overvalued in all likelihood.Therefore,it is vitally necessary to carry out researches on the effects of environmental change on runoff process of snowmelt flood,design flood and flood control uncertainty by reservoir regulation,explore the frequency calculation method of inconsistent snowmelt flood,and evaluate the effect of environmental change quantitatively.Ken Swart Reservoir Catchment of Manas River was set as the study region in this paper.Based on variation diagnosis,the contribution of climate change and human activities to the runoff process of snowmelt flood was studied.Quantitative effects of environmental change on reservoir design flood and flood control uncertainty by reservoir regulation were given on the basis of frequency calculation and analysis of inconsistent snowmelt flood,which can provide technical support and scientific guidance for design flood revision,planning and management of regional flood control and flood resources safe utilization under changing environment.The main research contents and results are as follows:(1)Adopting the methods of Pettitt test and Mann-Kendall test to analyze the variation and trend characteristics of hydrological element sequences,respectively,and the change points of average annual temperature,annual evaporation and annual runoff series were most likely to be 1979,1996 and 1993,while the annual precipitation series did not appear change point.According to the characteristics of climate change and remote sensing data of different time phase,combining with hydrologic physical mechanism,the underlying surface of the catchment did not change much before and after 1990,so the variation of annual runoff was primarily due to climate change,and the change point was 1993.(2)Through inconsistency test of snowmelt flood characteristic sequences,the change points of annual maximum flood peak and annual maximum flood volume sequences were both 1993,combining with remote-sensing image data of different time phase,it could be judged comprehensively that climate change was the major cause of the variation of snowmelt flood characteristic sequences.The change point of temperature sequences was 1979,and antecedent precipitation sequences of 1,7,15,30 days got no change point,while the variation of antecedent precipitation sequences of 3,5 days occurred in 1971.Correlation analysis method was used to analyze the relationship between meteorological impact factors and snowmelt flood characteristic sequences,and temperature sequences,antecedent precipitation sequences of 1 day and 3 days were selected as the impact factors of snowmelt flood characteristic sequences for their strongest correlation.(3)By using the methods of linear regression,Mann-Kendall test and cumulative anomaly,the change and variation features of snowmelt flood runoff and its climate impact factors were analyzed.It turned out that snowmelt flood runoff,precipitation and evaporation showed the same increase trend,and the change point of snowmelt flood runoff was 1995.The contribution rates of precipitation,evaporation and human activities to the increase of snowmelt flood runoff during 1996-2010 were 59.64%,31.83% and 8.53%,respectively,by adopting methods of statistical analysis and slope change rate comparison of cumulant,which indicated the effect of climate change on snowmelt flood runoff was greater than that of human activities.(4)The calculation results of snowmelt flood characteristic sequences were compared and analyzed based on the theoretical consistent GAMLSS model and the inconsistent GAMLSS model with time and climate factors as the covariate,respectively.The results showed that the fitting residual of the consistent model did not satisfy the requirement of normal distribution,which indicated that the snowmelt flood characteristic sequences did not accord with the consistency assumption.The fitting residual of the inconsistent model with time as the covariate complied well with normal distribution,and the snowmelt flood characteristic sequences showed a rising tendency with the passage of time.The physical mechanism of the inconsistent model with climate factors as the covariate whose fitting residual satisfied general requirement of normal distribution accorded with objective reality and its quantile gray-scale maps well described the dynamic change process of snowmelt flood characteristic sequences under the influence of environmental change.(5)Models of mixed distribution and conditional probability distribution were analyzed and compared,and the inconsistent snowmelt flood characteristic sequences were calculated,it turned out that the fitting effect of the mixed distribution model were better than that of the conditional probability distribution.The design flood results of the mixed distribution model were compared with the results approved in 2008 and the results of the original sequences fitting with P-III distribution without considering variation.The results showed that under different design standards,the design results of snowmelt flood fitting with mixed distribution and P-III distribution were smaller then that approved in 2008,and the results fitting with mixed distribution were bigger than that fitting with traditional P-III distribution.(6)The snowmelt flood characteristic sequences were revised for consistency in terms of the decomposition-composition theory,and the parameter uncertainty fitting with P-III distribution for measured,backward revised and forward revised snowmelt flood characteristic sequences was estimated using Bayes theory and Gibbs-MCMC algorithm,and the evaluation in regard to the superiority of prediction interval was also carried out.The results showed that the influence of parameter uncertainty on calculation could be reduced through backward or forward revision for inconsistent snowmelt flood characteristic sequences,so as to improve the reliability of the prediction interval.Taking design flood derived from snowmelt flood characteristic sequences after consistent revision under the two conditions as inflow snowmelt flood to execute reservoir flood routing,the extreme risk rates of reservoir flood control were rechecked by frequency analysis method under the two conditions.Moreover,fuzzy risk analysis model of reservoir overtopping was constructed based on right trapezoidal fuzzy number to identify index intervals of fuzzy risk of reservoir overtopping under the past and current conditions,and the interval of fuzzy risk rates and the lower and upper limit values of absolute error of different significant level ? were estimated.The results showed that snowmelt flood was affected by the obvious increase of temperature in the catchment under current condition,which increases the risk of reservoir failure.The results can provide scientific reference for safe reservoir operation and safe utilization of flood resources. |
PDF Full Text Request