小流域泥石流启动降雨阈值预测与虹吸分流防治方法

发布时间：2018-01-14 23:34

本文关键词：小流域泥石流启动降雨阈值预测与虹吸分流防治方法　出处：《浙江大学》2017年博士论文　论文类型：学位论文

【摘要】：小流域具有面积小、沟谷长度短和汇流速度快等特点,降雨径流反应迅速,短历时高强度降雨在较短时间内就会使沟谷产生超强的水动力条件,诱发泥石流灾害。尤其是我国东南沿海地区,经常受台风暴雨影响,常常在许多小流域同时出现泥石流灾害。针对小流域泥石流灾害的特点,依托典型小流域的降雨量、径流量监测,探究不同降雨过程条件下径流量的变化规律,提出泥石流启动降雨阈值预测模型,创建泥石流防治的虹吸分流新方法。研究取得的主要进展如下:(1)选择小流域开展了降雨径流量监测,监测结果显示径流量曲线波动趋势与降雨量分布基本一致,降雨特征决定了径流产生能力。选取降雨量P、平均雨强IAVE、最大30min雨强I30、最大10min雨强I10、初始径流量Q初,降雨历时D作为降雨特征参数进行研究。利用灰色关联分析法求解降雨特征对径流产生能力的影响系数,计算结果显示平均最大30 min雨强I30对径流产生能力贡献最大。(2)降雨特征值与径流产生能力存在较强的非线性关系,以降雨量P、平均雨强IAVE、最大30 min雨强I30、最大10 min雨强I10、初始径流量Q初作为模型输入因子,运用移动最小二乘(MLS)模型,遗传神经网络(GABP)模型以及遗传支持向量机(GASVM)模型对径流变化系数进行预测,计算结果显示MLS模型预测精度最高,可以对径流产生能力进行预测。同时利用NAM水文模型计算沟谷径流量,计算结果显示NAM水文模型可以用于小流域沟谷径流量的预测。(3)考虑泥石流形成区下垫层对泥石流形成的影响,提出基于临界流量和基于临界流深的泥石流启动降雨阈值预测模型。两种模型都判定研究区小流域在两年重现期降雨条件下发生泥石流,判定结果与实际情况相符。对比两种预测模型,基于临界流深的泥石流启动降雨阈值模型不仅可以判断泥石流的发生,还可判断泥石流的发生强度以及范围。(4)提出了MLS泥石流平均流速预测模型,预测模型根据影响因素数值所在范围不同而应用不同的映射拟合函数,且通过若干个公式分段表达泥石流流速,并引入紧支概念捕捉影响因素数值的微小变化,从而获得了更高的预测精度。(5)创建了泥石流防治的虹吸分流新方法,修筑了虹吸分流泥石流防治试验工程。监测数据表明:虹吸分流技术可将沟谷内大部分洪水进行分离。经FL0-2D模型计算可知经分流作用后,下游形成区沟谷的最大径流深度明显减少,抑制了泥石流的发生。
[Abstract]:The small watershed has the characteristics of small area, short valley length and fast confluence speed, and the response of rainfall runoff is rapid. The short duration and high intensity rainfall will make the valley produce super hydrodynamic condition in a short time. Induced debris flow disasters, especially in the southeast coastal areas of China, often affected by typhoon rainstorms, often occur in many small watershed debris flow disasters. Aiming at the characteristics of debris flow disasters in small watershed. Based on the monitoring of rainfall and runoff in typical small watershed, this paper probes into the variation law of runoff under different rainfall process, and puts forward a prediction model of threshold of debris flow starting rainfall. A new siphon diversion method for debris flow prevention is established. The main progress of the research is as follows: 1) the rainfall runoff monitoring is carried out in a small watershed. The monitoring results show that the fluctuation trend of runoff curve is basically consistent with the rainfall distribution, and rainfall characteristics determine the runoff production capacity. Select rainfall P, average rainfall intensity IAVE, maximum 30min rain intensity I30. The maximum 10min rainfall intensity I10, the initial runoff Q and the rainfall duration D were studied as the rainfall characteristic parameters. The grey correlation analysis was used to solve the influence coefficient of rainfall characteristics on runoff production capacity. The calculated results show that the average maximum 30 min rain intensity I30 has the largest contribution to runoff generation capacity.) the rainfall characteristic value has a strong nonlinear relationship with runoff production capacity. The average rainfall intensity is IAVE.The maximum rainfall intensity is 30 min, the maximum 10 min rain intensity is I10, and the initial runoff Q is taken as the input factor of the model, and the moving least squares (MLS) model is used. The genetic neural network (MLS) model and the genetic support vector machine (GASVM) model are used to predict the runoff variation coefficient. The calculated results show that the MLS model has the highest prediction accuracy. The runoff generation capacity can be predicted, and the NAM hydrological model is used to calculate the runoff of the gully and valley. The results show that the NAM hydrological model can be used to predict the runoff of gully in small watershed.) the influence of the underlying layer of debris flow formation area on the formation of debris flow can be considered. A prediction model of debris flow initiation rainfall threshold based on critical flow and critical flow depth is proposed. Both models determine the occurrence of debris flow in a small watershed in the study area under the condition of two years of recurrent rainfall. Compared with the two prediction models, the threshold model of debris flow initiation rainfall based on critical flow depth can not only judge the occurrence of debris flow. The prediction model of average velocity of MLS debris flow is put forward. According to the value range of influencing factors, different mapping fitting functions are used in the prediction model. The velocity of debris flow is expressed by several formulas, and the concept of tight support is introduced to capture the small change of the value of influencing factors. Thus, a higher prediction accuracy. 5) has been established to create a new siphon diversion method for debris flow prevention. The siphon distributary debris flow prevention test project was constructed. The monitoring data show that siphon distributary technology can separate most of the flood in the gully. The FL0-2D model calculation shows that the siphon distributary effect can be obtained. The maximum runoff depth of the valley in the downstream formation area is obviously reduced, which inhibits the occurrence of debris flow.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P642.23

【相似文献】