两种常用小流域洪水计算方法的灵敏度分析

发布时间：2018-07-25 16:36

【摘要】：小流域的水文站点分布密度低,常常缺乏足够的水文信息和数据。有些小流域甚至缺乏基础雨量资料,但小流域洪水应用极为广泛且影响计算结果的经验参数较多,所以大部分小流域洪水计算公式都是经过一系列的假定和模型概化而建立的,这样使得计算参数大幅减少,同时又不至于计算结果偏差太大。虽然计算参数的大幅减少使计算方法更为简便,但仅有的几个计算参数对计算结果的影响也必然随之增大。本文介绍了两种常用的小流域洪水洪峰流量的计算方法,即水科院推理公式法和林平一法,并利用EHP软件各进行了三个实际算例的计算。通过对两种方法的主要计算参数的分析,筛选出不确定性较大的参数,对筛选出的参数进行±0.5%、±1%、±2%、±5%、±10%、±15%和±20%等7个不同幅度的不重复扰动,来计算参数的灵敏度系数大小。通过灵敏度系数的比较,即可得知各参数具有的误差对洪峰流量计算成果精度的影响程度。文中不仅分别对水科院推理公式法和林平一法中各自的参数进行了灵敏度分析,还对两种方法中涉及的共性参数进行了对比分析,以便得出其一般性的趋势。取得的主要研究结果如下： 1.推理公式法中损失参数μ、汇流参数m、河道纵比降J和24h最大降雨量H24等四个参数的灵敏度系数的绝对值大小依次为其中|QJ|和|Qμ|两个较小,均小于0.5,表明损失参数μ和河道纵比降J两个参数的灵敏度系数较小,对洪峰流量计算成果精度影响较小；其次,影响较大的参数为汇流参数m,|Qm|主要分布在0.8～1.2之间,对洪峰流量的计算成果影响较大；影响最大的为|QH24h|,并且大于1,表明在上述四个参数中24h最大降雨量H24灵敏度系数最大。另外,各参数在5%范围以内变化时,其灵敏度系数随着参数的扰动的波动较大。 2.林平一法中稳定下渗率μ和河道坡度J。两个参数的灵敏度系数的绝对值|q|和|OJc|较小,其绝对值均小于0.5,表明稳定下渗率μ和河道坡度Jc两个参数对洪峰流量计算成果影响较小；河道汇流糙率Nc的灵敏度系数的绝对值|QNc|一般小于1,但接近于1,表明河道汇流糙率Nc对洪峰流量计算成果影响较大；其中灵敏度最大的参数为24h最大降雨量H24,其灵敏度系数的绝对值|QH24h|大于1,表明在上述四个参数中24h最大降雨量H24灵敏度系数最大。另外,各参数在5%范围以内变化时,其灵敏度系数随着参数的扰动的波动较大。 3.推理公式法和林平一法两种方法的共性参数主要有J（c）和H24h。参数J（c）的灵敏度系数0QJ（c）1，表明两种方法中的J（c）与其灵敏度系数QJ（c）均呈正相关，并且参数J（c）的变化对洪峰流量产生的误差起到缩小作用；随着参数J（c）的增大，QJ（c）一般呈减小趋势；相比较而言，参数J（c）的灵敏度系数QJ（c）在推理公式法中比林平一法中更大一些。参数H24h的灵敏度系数QH241，，表明两种方法中的H24h与其灵敏度系数QH24均呈正相关，并且参数H24h的变化对洪峰流量产生的误差起到扩大作用；随着参数H24h的增大，QH24一般呈增大趋势；参数H24h的其灵敏度系数QH24的大小关系在推理公式法中和林平一法中无明显的比较规律。
[Abstract]:The distribution density of hydrological stations in small basins is low and often lacks sufficient hydrological information and data. Some small basins even lack the basic rainfall data, but the flood application of small watershed is very extensive and the empirical parameters affecting the calculation results are many, so most of the flood calculation public formulas in most small basins are generalized by a series of assumptions and model generalizability. In this way, the calculation parameters are greatly reduced and the deviation of the calculation results is not too large. Although the large reduction of the calculation parameters makes the calculation more convenient, the only several calculation parameters will inevitably increase the effect of the calculation results.
This paper introduces two methods for calculating flood and flood peak flow of small watersheds, namely, the method of reasoning formula of the Academy of water science and the Lin Ping one method, and the calculation of three practical examples using EHP software. Through the analysis of the main calculation parameters of the two methods, the uncertain parameters are screened out, and the selected parameters are 0.5% and +. 1%, + 2%, + 5%, + 10%, + 15% and + 20% and other 7 different amplitude non repetitive disturbances, to calculate the sensitivity coefficient of the parameters. Through the comparison of the sensitivity coefficient, we can know the degree of influence of the error on the accuracy of the calculation results of the flood peak flow. The sensitivity analysis of the parameters is carried out, and the common parameters involved in the two methods are compared and analyzed in order to get the general trend. The main results obtained are as follows:
1. the absolute value of the sensitivity coefficients of four parameters, such as the loss parameter mu, the confluence parameter m, the channel longitudinal ratio drop J and the 24h maximum rainfall H24, are respectively |QJ| and |Q Mu two smaller, both less than 0.5, indicating that the loss parameter mu and the channel longitudinal ratio drop J two parameters are smaller, and the calculation results of the flood peak flow are refined. Second, the parameter of large influence is m, and |Qm| is mainly distributed between 0.8 and 1.2, which has great influence on the calculation results of peak flow; the maximum impact is |QH24h|, and more than 1, which indicates that the maximum 24h precipitation sensitivity coefficient of the maximum rainfall is maximum in the above four parameters. In addition, when the parameters vary within the range of 5%, The sensitivity coefficient fluctuates with the perturbation of the parameters.
2. the absolute value |q| and |OJc| of the sensitivity coefficient of the stable infiltration rate and the channel gradient J. two parameters are smaller than 0.5. It shows that the steady infiltration rate Mu and the channel slope Jc two parameters have little influence on the calculation results of the flood peak flow, and the absolute value of the sensitivity coefficient of the channel flow roughness Nc is generally less than 1, But close to 1, it shows that the river flow roughness Nc has great influence on the calculation results of the flood peak flow. The maximum sensitivity parameter is 24h maximum rainfall H24, and the absolute value of the sensitivity coefficient is more than 1, indicating that the maximum 24h rainfall H24 sensitivity coefficient is maximum in the four parameters of the above four parameters. In addition, when each parameter varies within 5% range, The sensitivity coefficient fluctuates with the perturbation of the parameters.
3. the common parameters of 3. reasoning formula method and two methods are mainly J (c) and H24h. parameter J (c) sensitivity coefficient 0QJ (c) 1. It shows that J (c) in the two methods is positively correlated with the sensitivity coefficient QJ (c), and the variation of the parameter J is reduced to the error of the peak flow rate. In comparison, the sensitivity coefficient QJ (c) of parameter J (c) is larger in the reasoning formula method than in the Linping method. The sensitivity coefficient QH241 of the parameter H24h shows that the H24h of the two methods is positively correlated with the sensitivity coefficient QH24, and the variation of the parameter H24h plays an important role in the error of the peak flow. With the increase of parameter H24h, QH24 generally assumes an increasing trend, and the relation between the sensitivity coefficient QH24 of parameter H24h has no obvious comparative law in the inference formula method and Lin Ping one method.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TV122

【参考文献】