耦合ECMWF降雨集合预报的水库优化调度研究

发布时间：2018-03-16 11:47

本文选题：降雨集合预报　切入点：评分方法　出处：《大连理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：提高水资源利用效率,是缓解日益加剧的水资源危机的重要措施之一。提高水库的调度决策水平,能有效提高水能资源利用率,增加发电效益,促进国民经济的发展。在水库的优化调度中,径流的未知和预报的不确定性是水库调度面临的最大的难题。传统的“单一”降雨数值预报的不确定性较大,径流预报不够准确,可能导致调度出现较大偏差。为了充分考虑降雨预报的不确定性,本文使用欧洲中期天气预报中心(ECMWF)发布的降雨集合预报信息建立径流描述模型,进行中期径流预报。然后利用径流预报信息建立桓仁水库发电优化调度模型,进行水库的优化调度研究。本文研究成果主要有以下几个方面： (1)通过TS评分、BS评分和Talagrand分布对ECMWF的集合预报系统进行检验评价,结果显示：随着预见期的延长及降雨量级的提高,ECMWF的集合预报精度逐渐降低,但是集合平均值预报效果比控制预报的好；ECMWF集合预报系统的短期降雨预报集合成员不够发散,而中长期降雨预报存在着系统偏差。 (2)对降雨集合预报的精度进行分析,结果显示：降雨集合预报的空报率大于漏报率,并且随着预见期的延长和降雨量级的提高,准确率逐渐降低；无雨预报的准确率较高,小雨和中雨预报的准确率也可以接受。降雨集合预报信息达到一定精度,可以应用在径流预报中。 (3)针对桓仁水库以上流域汛期和非汛期流域产汇流的特点,本文在汛期和非汛期分别建立径流模型。采用ECMWF发布的未来5d和10d的降雨集合预报信息输入已建立的模型,进行径流集合预报。结果显示：5d降雨集合预报的径流预报效果较好；而10d降雨集合预报的径流过程出现预报偏大现象,由此可见,降雨集合预报信息的预报效果随着预见期的延长有所降低。 (4)将ECMWF降雨集合预报所得径流预报信息分别采用控制预报、集合平均值和区间分级的方式输入随机动态规划模型进行水库发电调度,结果显示：应用以5d为时段的降雨集合预报信息进行水库优化调度,可以提高水资源利用率,增加水库发电的经济效益；由于预见期的延长,降雨集合预报精度降低,10d的降雨集合预报甚至不能直接应用在调度中。
[Abstract]:Improving the efficiency of water resources utilization is one of the important measures to alleviate the worsening water resources crisis. Improving the decision-making level of reservoir dispatching can effectively improve the utilization ratio of hydropower resources and increase the power generation efficiency. To promote the development of national economy. In the optimal operation of reservoirs, the unknown runoff and the uncertainty of forecast are the most difficult problems faced by reservoir operation. The traditional "single" rainfall numerical forecast is uncertain, and the runoff forecast is not accurate enough. In order to fully consider the uncertainty of rainfall forecast, the runoff description model is established by using the rainfall ensemble forecast information published by ECMWF, the European Center for Medium-Term Weather forecasting (ECMWF). Using runoff forecast information to establish Huanren reservoir power generation optimal operation model, to study the reservoir optimal operation. The main research results of this paper are as follows:. 1) the ensemble prediction system of ECMWF is tested and evaluated by TS score and BS score and Talagrand distribution. The results show that the accuracy of ensemble prediction of ECMWF decreases gradually with the extension of forecast period and the increase of rainfall level. However, the ensemble average forecasting effect is less than that of the ECMWF ensemble forecasting system controlled by the control, and there is not enough divergence in the short-term rainfall forecast ensemble members, but there is a systematic deviation in the medium- and long-term rainfall forecast. 2) the accuracy of rainfall ensemble forecast is analyzed. The results show that the accuracy rate of rainfall ensemble forecast is higher than that of missing report rate, and with the extension of forecast period and the increase of rainfall level, the accuracy of rainfall ensemble forecast decreases gradually, and the accuracy of rain free forecast is higher. The accuracy of light rain and moderate rain forecasting is acceptable, and the rainfall aggregate forecast information can be used in runoff forecast. 3) according to the characteristics of flood season and non-flood season, runoff models are established in flood season and non-flood season, respectively. The models of rainfall ensemble forecast for the next 5 days and 10 days published by ECMWF are used to input the established models. Runoff ensemble forecasting is carried out. The results show that the runoff forecasting effect of rainfall ensemble forecasting is better than that of 10 days rainfall ensemble forecast, and the runoff process of 10d rainfall ensemble forecast is larger than that of runoff ensemble forecast, so it can be seen, The forecast effect of rainfall ensemble forecast information decreases with the extension of forecast period. 4) the runoff forecast information from ECMWF rainfall ensemble forecast is inputted into stochastic dynamic programming model for reservoir power generation by means of control forecast, set average value and interval classification, respectively. The results show that the optimal operation of reservoirs with rainfall aggregate forecast information of 5 days can improve the utilization of water resources and increase the economic benefits of power generation. The rainfall ensemble forecast with 10 days lower precision of rainfall ensemble forecast can not even be directly applied in the dispatching.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TV697.1;P338

【参考文献】