Optimization method for monthly runoff stochastic simulation sequence
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
Hydrologic stochastic model can provide a scientific and decision-making basis for alleviating water resources problems by simulating a large number of random sequences for hydrologic system.However,the simulation series generated by the same stochastic model was quite different in time and process changes and the traditional statistical parameter test method ignored the variation characteristics of runoff as a time series,which can not fully reflect the complex variation characteristics of flow in time series distribution,and failed to select the most suitable runoff simulation series for engineering planning and design and risk analysis.Therefore,a set of evaluation index system of monthly runoff random generation series is proposed,which incoporates statistical characteristics of runoff simulation series at different time intersections and the complex variation characteristics of runoff series.Three hydrological stations in the main and tributary streams of the Huaihe River basin are taken as the research object to establish a seasonal autoregressive model,three evaluation methods are selected based on constructed index system and multiple sets of randomly generated monthly runoff simulation sequences.The sequence comparison aimed to provide a more accurate and reliable new way to identify the random simulation sequence of monthly runoff. Starting from the nonlinear and complex characteristics of runoff series,indicators such as concentration,uneven coefficient and sample entropy based on traditional test indicators,and a set of index systems is proposed that characterize the statistical parameter characteristics and complex changes of the runoff series.The accuracy of the simulation sequence was analyzed,and the simulation sequence was optimized by the gray correlation analysis method.The fitting results of the three evaluation methods (traditional index,minimum entropy index of sample and index) were compared and analyzed to illustrate the rationality of the index.Moreover,the advantages of the proposed evaluation index system were further verified from the two aspects of the abnormal situation of the year allocation and the difference of annual time history distribution (process) of runoff. Comparison results of the optimized sequences of the three evaluation methods showed that:(1) In the evaluation of statistical characteristics of runoff simulation series at different time intersections,the difference of the average absolute percentage error of the simulation sequence optimized by the three indexes was small,and the maximum error was 2%.(2) In the evaluation of complex change characteristics,the difference of average absolute percentage error was large.The maximum error difference between the optimization sequence based on the index and the traditional index was 8%,and the maximum error difference between the optimization sequence based on the index and the sample entropy index was 4%.The selected sequence performed well in the evaluation of statistical parameter characteristics and complex change characteristics,and the most significant difference was in complex change characteristics compared with the other two evaluation methods.(3) The results showed that the performance of the series selected was good in the annual distribution characteristics of each level,and the relative error was generally smaller than that of the series selected by traditional index and sample entropy index,and the difference between the relative error in normal and dry years was the most obvious. In conclusion,the three evaluation methods were effective in maintaining the characteristics of the runoff series,but the previous index system can not comprehensively evaluate the simulation series,which may omit the important details of the runoff series,thus affecting the accuracy of the simulation series.In this index system,the indexes which represented the complex change characteristics can simply and effectively distinguish the ability of simulation series in maintaining the complex change characteristics of measured series.The selected series can better reflect the change process and complex characteristics of the measured series,and the overall fitting effect of annual distribution was also the best.The order of merit and demerit of the three methods was the proposed index of this paper>the index of sample entropy>the traditional index.The results showed that the proposed index system comprehensively considered the statistical parameters of the section and the characteristics of complex changes was feasible.The abnormal situation of the year allocation and the difference of annual time history distribution (process) of runoff were analyzed,the results showed that the advantage of proposed index,which is beneficial to a comprehensive evaluation of runoff simulation series and realizes the random simulation comparison and optimization among different series,thereby provided more accurate hydrological simulation sequences.
Keywords:
Project Supported:
Clc Number:
Mobile website
