滦河潘家口水库控制流域干旱等级预测研究

发布时间：2018-09-10 20:36

【摘要】：随着全球气候的变化,以及社会经济发展带来的用水需求的增加,滦河流域水资源短缺和干旱问题日益突出,一直制约着其社会经济的发展。为此,本文从气象干旱和水文干旱研究的角度出发,对滦河潘家口水库控制流域的干旱等级预测方法进行了研究,构建了三种预测模型,以期找到更好的预测方法、预测模型提高流域干旱等级预测的精度,为流域干旱应对和水资源科学管理提供参考,主要研究内容和成果如下:(1)干旱等级评价模式的建立。基于滦河潘家口水库控制流域的水文资料计算获得标准化降水指数(SPI)、标准化径流指数(SRI)时间序列,再参照干旱等级的划分标准,分别获得对应的SPI、SRI等级序列,即气象、水文干旱等级序列,并采用历史干旱年份的旱情进行了验证,评价结果与实际旱情基本一致。(2)预测方法的分析与预测模型的构建。以SPI和SRI时间序列为研究对象,分别建立了三维对数线性模型和加权马尔可夫链模型,并首次运用混沌时间序列Volterra自适应滤波器模型,对未来SPI和SRI等级进行预测模拟,实现了短期和中长期干旱等级的预测。(3)预测模型的应用与预测精度分析。对不同预测模型的预测精度进行分析,结果表明,在滦河潘家口水库控制流域干旱等级预测中,三种预测模型各有其优势与不足,三维对数线性预测法适用于预见期为1个月的干旱等级预测,模型的预测精度随预测步长的增加而下降,不能用于较长时间的预测;加权马尔可夫链预测模型对干旱的预测准确率由高到低依次为无旱、中旱、重旱/特旱、轻旱,且预测精度与干旱等级所处的稳定状态有显著关系,当干旱等级的发展过程相对平稳时其预测能力较强,而当干旱等级剧变时其预测能力较弱;将相空间重构技术、Volterra级数展开技术以及自适应滤波器优化技术相结合,提出了一种新的干旱等级预测方法,模型适用于滦河潘家口水库流域中长期干旱等级预测,且具有较高的预测精度,同时,干旱等级预测的预见期增长至12个月,可为水资源系统的中长期决策或规划提供更多的信息支持。
[Abstract]:With the change of global climate and the increase of water demand brought by the development of social economy, the problems of water resources shortage and drought in Luanhe River Basin are becoming increasingly prominent, which has been restricting the development of its social economy. Therefore, from the point of view of the study of meteorological drought and hydrological drought, this paper studies the prediction methods of drought grade in Panjiakou reservoir control basin of Luanhe River, and constructs three forecasting models in order to find a better prediction method. The prediction model improves the precision of drought grade prediction and provides a reference for drought response and scientific management of water resources. The main research contents and results are as follows: (1) Establishment of drought grade evaluation model. Based on the hydrological data of Panjiakou Reservoir control basin of Luanhe River, the standardized precipitation index (SPI), standardized runoff index (SRI) time series is obtained, and the corresponding SPI,SRI grade series, that is, meteorology, is obtained according to the criteria of drought classification. The hydrological drought grade series is verified by the drought condition in the historical drought year, and the evaluation results are basically consistent with the actual drought situation. (2) the analysis of forecasting method and the construction of forecasting model. Taking SPI and SRI time series as research objects, the three dimensional logarithmic linear model and weighted Markov chain model are established, respectively. The chaotic time series Volterra adaptive filter model is used for the first time to predict and simulate the future SPI and SRI grades. The prediction of drought grade in short and long term is realized. (3) the application of forecasting model and the analysis of prediction precision. The prediction accuracy of different forecasting models is analyzed. The results show that the three prediction models have their own advantages and disadvantages in the prediction of the drought grade of the Panjiakou Reservoir in Luanhe River, which is controlled by the Panjiakou Reservoir. The three dimensional logarithmic linear prediction method is suitable for drought grade prediction with a predicted period of one month. The prediction accuracy of the model decreases with the increase of the prediction step and cannot be used for longer prediction. The prediction accuracy of the weighted Markov chain model for drought prediction is from high to low in the order of no drought, moderate drought, severe drought / special drought, and light drought, and the prediction accuracy is significantly related to the stable state of drought grade. When the development process of drought grade is relatively stable, its prediction ability is stronger, but when the drought grade changes rapidly, its prediction ability is weak, which combines the phase space reconstruction technology with Volterra series expansion technology and adaptive filter optimization technology. A new method of drought grade prediction is put forward. The model is suitable for long term drought grade prediction in Panjiakou reservoir basin of Luanhe River, and has high prediction accuracy. At the same time, the forecast period of drought grade prediction increases to 12 months. It can provide more information support for medium-and long-term decision-making or planning of water resources system.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TV697;P426.616

【相似文献】