基于SWAT模型的大汶河流域不同分辨率数据组合下地表径流的模拟研究

发布时间：2018-04-18 03:17

本文选题：大汶河流域 + SWAT模型　；参考：《山东农业大学》2017年硕士论文

【摘要】：随着水资源管理问题的日益严峻,利用水文模型模拟流域等水循环过程是解决水资源管理问题行之有效的途径。本文借助SWAT模型,以大汶河流域为研究区,采用SUFI-2方法进行参数的率定,以Nash-Sutcliffe为目标函数,并开发了便于输入数据预处理的《图像预处理系统》,分别对2008-2009年大汶河流域8组不同分辨率组合的数据(DEM数据和土地利用数据)进行了16次迭代,每次迭代模拟300次。最后本研究对2010-2011年的径流数据进行了验证(DEM数据和土地利用数据分辨率均为30m),主要的研究结果包括以下几个方面:(1)本研究对六组分辨率匹配组合敏感性最强的6个参数进行了分析,得出各组合下敏感性最强的6个参数相同,分别是SCS径流曲线数(CN2)、土壤湿容重(SOL_BD)、土壤饱和渗透系数(SOL_K)、平均坡降(SLSUBBSN)、平均比降(HRU_SLP)、河岸调蓄的基流α因子(ALPHA_BNK),说明不同分辨率组合对于模型模拟参数的敏感性强弱影响并不大。6个参数的最优值在不同分辨率组合下的结果不同,说明不同分辨率组合下对参数的最优值影响比较大。(2)本研究分别对八组不同分辨率组合下的高程、流域面积、河网结构、子流域特征、HRU数、径流特征进行了分析。结果表明,高程、流域面积、河网结构三个特征因均由DEM数据提取获得,在相同来源、相同数据精度的DEM数据下,不同分辨率组合的模拟对这三个流域特征影响不大,各组合下三个特征值基本相同,但对子流域特征、HRU数、河道长度、径流量有一定的影响。在分辨率匹配组合中,随着各组合分辨率的降低,子流域数目、HRU数、河道长度以及径流量均呈下降趋势,其中30*30分辨率组合下模拟的河道长度、年均径流量最为接近实测数据,其中河道长度偏差仅为-1.07%;在分辨率不匹配组合中,土地利用分辨率高的组合HRU数更多,HRU是SWAT水文模型水文计算的基本单元,相应的也影响着径流量,而子流域特征以及河道的长度并不随分辨率组合的变化而改变,这一实验结果表明,子流域特征以及河道的长度受DEM分辨率的影响,而不受土地利用数据分辨率的影响。(3)本研究在不同分辨率组合下模型校准期模拟地表径流以及30m分辨率DEM和土地利用数据组合验证期模拟的地表径流均得到了较为满意的模拟结果。研究通过添加两组分辨率不匹配的组合进行模拟,得出在数据分辨率不匹配的情况下,SWAT模型会自动将数据统一分辨率,同最高分辨率的数据,但是模型输入空间数据分辨率不匹配时模拟年均径流量与分辨率匹配时模拟的结果有一定的差异。这说明在模型模拟前统一输入空间数据的分辨率是有必要的。
[Abstract]:With the problem of water resources management becoming more and more serious, it is an effective way to use hydrological model to simulate the process of water circulation in watersheds.In this paper, with the help of SWAT model and the Davin River basin as the study area, the rate of parameters is determined by SUFI-2 method, and the objective function is Nash-Sutcliffe.An image preprocessing system, which is convenient for input data preprocessing, was developed, and 16 iterations were carried out for 8 groups of Davin River basin data (Dem data and land use data with different resolutions) from 2008 to 2009. Each iteration was simulated 300 times.Finally, this study verified that the resolution of both Dem data and land use data was 30 mg. The main results included the following aspects: 1) the most sensitive group of six groups of resolution matching combinations.Three parameters are analyzed,The six most sensitive parameters of each combination are the same.They are SCS runoff curve, soil wet bulk density, soil saturation permeability coefficient, average slope gradient, average slope gradient, average specific gradient, average specific drop, 伪 factor of basic flow and ALPHABNKN, which indicate the sensitivity of different resolution combinations to the parameters of model simulation.The optimal values of the six parameters are different at different resolution combinations.In this study, the elevation, basin area, river network structure, subbasin characteristics and runoff characteristics of eight groups of combinations with different resolution were analyzed respectively.The results show that the elevation, basin area and river network structure are all extracted from DEM data. Under the same DEM data of the same source and the same data precision, the simulation of different resolution combinations has little effect on the three watershed features.The three eigenvalues of each combination are basically the same, but they have a certain effect on the number of HRU, the length of the channel and the runoff of the subbasin.In the resolution matching combination, with the decrease of the resolution of each combination, the number of HRU, the length of the river channel and the runoff of the sub-watershed are all decreasing. The average annual runoff of the simulated channel length under the combination of 30 ~ 30 resolution is the closest to the measured data.Among them, the deviation of channel length is only -1.07.The HRU number of combination with high land use resolution is the basic unit of hydrological calculation of SWAT hydrological model, and it also affects the runoff, in the combination of mismatch resolution, the combination with high land use resolution is the basic unit for hydrological calculation of SWAT hydrological model.However, the characteristics of subbasin and the length of channel do not change with the change of resolution combination. The experimental results show that the characteristics of subbasin and the length of channel are affected by DEM resolution.But not affected by the resolution of land use data. (3) in this study, the surface runoff of model calibration period and 30 m resolution DEM and land use data combination verification period were compared with each other under different resolution combinations.For satisfactory simulation results.By adding two sets of resolution mismatch combinations to simulate, it is concluded that the SWAT model will automatically unify the data with the highest resolution when the data resolution is not matched.However, when the spatial data resolution of the model is not matched, there is a certain difference between the simulation results when the average annual runoff is matched with the resolution.This shows that it is necessary to unify the resolution of input spatial data before model simulation.
【学位授予单位】：山东农业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P333;TP391.41

【相似文献】