诺敏河流域径流变化规律分析及预报方法研究

发布时间：2018-05-06 16:46

本文选题：径流 + 变化特征　；参考：《东北农业大学》2014年硕士论文

【摘要】：在自然界中,无论对于生物体的结构组成、生命活动,还是生态系统,水资源都起着至关重要的作用,是不可替代的资源。而我国的水资源问题已经十分突出,尤其是水资源短缺、水环境污染以及旱涝灾害问题,严重影响着我国社会经济的发展,成为其重要的制约因素。作为水资源最主要的来源之一,河川径流是水资源合理开发利用、优化配置的重要依据。在整个水文循环的系统中,径流的变化起着主导作用。如何准确分析河川径流的变化规律及未来发展趋势,科学合理地利用现有水资源成为最为迫切的问题。深入研究流域水资源的变化规律,并对其未来的发展趋势进行准确的预测,对流域合理开发利用水资源具有重要意义。本论文立足于嫩江支流诺敏河,研究了流域径流的变化规律和未来发展趋势问题。通过实地调研与数学模型相结合,先采用数理统计的方法分析了流域径流的基本统计特征、年内分配和年际变化规律,接着依次分析了流域径流的趋势、周期和突变等变化特征,各采用多种方法验证了规律的准确性与可靠性,然后应用BP神经网络和鱼群优化的BP神经网络预测了流域的月径流,应用EMD耦合谐波模型和基于EMD分解的AFSA-BP神经网络预测了流域的年径流的变化趋势,主要研究成果如下： (1)通过对流域径流量基本统计特征、年内分配不均匀性、集中程度、变化幅度的分析,以及年际变化的总体特征和距平分析等,得出年径流呈正偏分布,径流量分布较分散,径流量的年内分配不均,年际变化较大。流域径流年内变化幅度较大,整体呈现出波动上升又波动下降的趋势。年内分配曲线呈单峰形。径流主要集中在7月份和8月份,1-3月径流量最小。 (2)诺敏河流域年径流呈波动变化,增加和减少相互交替,整体上年径流量大致呈下降趋势,但趋势性不显著。年径流序列未来的趋势与过去相同,即未来有下降趋势。通过周期分析与多时间尺度分析得知,诺敏河流域年径流存在4年左右的短周期与30年左右的长周期。诺敏河流域年径流最明显的变异点发生在1998年,此外还存在1963年的突变。 (3)周期分析中,基于EMD的多时间尺度分析显示了径流序列变化的多时间尺度性、多层次性和复杂性,并通过计算各个分量的方差贡献率分析出主要的周期,优于单一的周期分析方法。突变分析中,Mann-Kendall突变检测法和Pettitt突变点检验法更全面一些。此外,Mann-Kendall秩次相关分析法和重标极差分析法可以应用到多个方面的规律分析中,具有更大的实用性。 (4)充分考虑各种数学模型的优缺点,将多种模型耦合在一起建立符合流域特点的动态预测模型,并对流域径流量进行预测。包括用AFSA-BP神经网络模型预测流域月径流,用EMD耦合谐波的方法和基于EMD的AFSA-BP神经网络模型预测年径流量。
[Abstract]:In nature, water resources play an important role in the composition of organisms, life activities and ecosystems, and are irreplaceable resources. However, the problems of water resources in China have become very prominent, especially the shortage of water resources, the pollution of water environment and the disaster of drought and waterlogging, which seriously affect the development of our country's social economy and become an important restrictive factor. As one of the most important sources of water resources, river runoff is an important basis for rational development and utilization of water resources. The variation of runoff plays a leading role in the whole hydrological cycle system. How to accurately analyze the variation law and future development trend of river runoff and make scientific and reasonable use of existing water resources has become the most urgent problem. It is of great significance for the rational exploitation and utilization of water resources in the basin to study the changing law of water resources and forecast the future development trend of water resources. Based on the Nenjiang tributaries, this paper studies the variation law and future development trend of runoff basin. Through the combination of field investigation and mathematical model, the basic statistical characteristics, distribution and interannual variation of watershed runoff are analyzed by mathematical statistics, and then the trend of watershed runoff is analyzed in turn. The accuracy and reliability of the rules are verified by various methods, and the monthly runoff of the watershed is predicted by BP neural network and optimized BP neural network. EMD coupled harmonic model and AFSA-BP neural network based on EMD decomposition are used to predict the trend of annual runoff. The main results are as follows: 1) through the analysis of the basic statistical characteristics of runoff in the watershed, the inhomogeneity of distribution in the year, the degree of concentration, the range of variation, the overall characteristics of interannual variation and the analysis of anomaly, it is concluded that the annual runoff is positively biased, and the distribution of runoff is scattered. The annual distribution of runoff is uneven and the interannual variation is great. The range of annual runoff variation is large, and the overall fluctuation is rising and decreasing. The distribution curve of the year presents a single peak. The runoff was mainly concentrated in July and August. 2) the annual runoff of the Normin River basin fluctuates, increases and decreases alternately, the overall annual runoff generally shows a downward trend, but the trend is not significant. The future trend of the annual runoff series is the same as that of the past, that is, there is a downward trend in the future. By means of periodic analysis and multi-time scale analysis, it is found that the annual runoff of the Normin River basin has a short period of about 4 years and a long period of about 30 years. The most obvious variation of annual runoff occurred in 1998, in addition to the 1963 mutation. In the periodic analysis, the multi-time scale analysis based on EMD shows the multi-time scale, multi-level and complexity of runoff series change, and the main period is analyzed by calculating the variance contribution rate of each component. It is superior to the single periodic analysis method. Mann-Kendall mutation detection and Pettitt mutation point test are more comprehensive in mutation analysis. In addition, Mann-Kendall rank correlation analysis and rescaling range analysis can be applied to many aspects of the law analysis, which is more practical. (4) considering the merits and demerits of various mathematical models, a dynamic forecasting model is established by coupling various mathematical models together, and the runoff of the basin is predicted. The AFSA-BP neural network model is used to predict the monthly runoff, the EMD coupled harmonic method and the AFSA-BP neural network model based on EMD to predict the annual runoff.
【学位授予单位】：东北农业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P333;P338

【参考文献】