【作者】 李辉；

【导师】 陈晓玲；

【作者基本信息】 武汉大学 ， 地图学与地理信息系统， 2007， 博士

【摘要】 土壤侵蚀是当今世界面临的主要问题之一。土壤侵蚀使生态环境恶化、土地生产力下降、耕地减少、污染水质、江河、湖泊淤塞、威胁城市和乡村防洪安全、加剧旱涝风沙等灾害。它使人类赖以生存并越来越短缺的土地资源遭受退化和损失。土壤侵蚀模型作为了解土壤侵蚀过程与强度、指导人们合理利用土地资源、管理和维持人类长期生存环境的重要技术工具,受到世界各国的普遍重视。按照模型的建模手段和方法,土壤侵蚀模型一般分为经验统计模型和物理过程模型。由于经验模型适用范围有限,因而限制了其推广应用;而物理模型反映了侵蚀的各个过程,具有较强的适应范围,是目前模型发展的趋势。由于侵蚀过程存在空间上的变异,所以物理模型通常是分布式的。尤其是进人到90年代以后,地理信息技术有了长足的发展,遥感(RS)和地理信息系统(GIS)的广泛应用,为数据的提取、贮存、处理和计算,提供了灵活、方便的手段,大大推动了分布式模型的发展及应用。我国是世界上土壤侵蚀最严重的国家之一,土壤侵蚀面积占全国总土地面积38.2%。每年因此流失的土壤达50亿吨以上、损失的土地面积6.67万公顷以上,大量氮、磷、钾肥料和多种微量元素随之付诸东流。目前国内用于预报土壤侵蚀的模型以经验模型为主,分布式模型不是很常见,而且土壤侵蚀模型的研究主要集中在北方的黄土高原地区,而广大南方地区也是我国水蚀最严重的区域之一,还未公开见到分布式土壤侵蚀模型的发表。因此迫切需要结合RS, GIS技术,构建适合于我国南方地区的分布式土壤侵蚀模型。本文在分析国内外分布式土壤侵蚀模型研究现状与存在问题的基础上,以土壤侵蚀力学、水力学、地貌学、泥沙运动学等学科的理论为基础,结合现代空间信息技术,初步研制开发了适合我国南方流域的、以次降雨为基础的、基于DEM的分布式土壤侵蚀模型。本文还讨论了模型参数率定的方法并利用实测数据对模型进行了率定和验证,最后使用该模型研究了躁口水流域的土壤侵蚀状况。本文的主要研究成果如下:1.基于DEM的流域参数提取。本文讨论了DEM中提取流域参数的方法,介绍了使用D8方法提取流向矩阵以及流量累积矩阵以及提取水系和流域边界的方法。本文研究了DEM预处理方法,并提出了基于三方向搜索的预处理算法。通过与ArcHydro中DEM处理方法比较后,结果表明,本文提出的方法在提取水系时能够消除在平坦区域出现的“平行线”现象;并且本文方法对DEM高程值的累计修改量也小于ArcHydro方法。对研究区水系提取的结果表明,该方法在水系提取精度上也要优于ArcHydro方法。2.分布式水文模拟研究。分布式土壤侵蚀模型一般都是建立在分布式水文模型基础之上的。本文探讨了研究区产流机制,研究了降雨、植被截留和土壤入渗等水文过程,并利用运动波和扩散波对次降雨产生的洪水逐栅格进行演进,获取各个栅格的水力要素,为侵蚀产沙模块提供基础。本文在研究不同坡度下运动波和扩散波的不同模拟精度基础上,提出了根据坡度来确定洪水演算模式的方法。本文采取了分级汇流的模式,对流域洪水从距离流域出口点最远处的分水岭开始,由远及近地进行分级演进模拟。通过研究,本文提出了一种有效存储分级汇流栅格拓扑结构的存储方式,该方法简单易行,能够准确地对流域进行分级演算,并且具有较高效率。3.侵蚀产沙模拟研究。在国外和国内黄土高原土壤侵蚀研究成果的基础上,本文分析了雨滴溅蚀、片蚀和细沟侵蚀等坡面侵蚀过程,及对坡面径流的搬运能力,并分别进行模拟。从而得到每个栅格的侵蚀和产沙量。然后利用泥沙连续方程和分级汇沙原理将各个栅格产沙量演算到流域出口,得到整个流域的次降雨产沙量。4.模型参数率定。运用计算机程序对流域洪水进行准确、快速地模拟的一个重要前提就是优化模型的参数,使计算结果更加准确。本文介绍了常用的参数优化方法,并根据分布式模型自身的特点,选用人机结合的方法对研究区的参数进行率定。5.模型验证与应用。本文以鄱阳湖流域修水水系上游的躁口水流域为研究对象,利用实测历史数据对模型进行验证。统计结果表明,该模型对流域产流产沙的模拟结果是可信的。在模型验证的基础上,使用模型来研究该流域的一场降雨引起的土壤侵蚀,结果表明坡度>15%的坡耕地是主要泥沙来源。更多还原

【Abstract】 Accelerated soil erosion has been globally recognized as a serious problem since people took up agriculture. It is becoming one of the most serious environmental problems in the world. Soil erosion affects soil productivity by changing soil properties, and particularly by destroying topsoil structure, reducing soil volume and water holding capacity, reducing infiltration, increasing run-off and washing away plant nutrients such as nitrogen, phosphorous, and organic matter. The resulting sediments themselves act as the carrier of pollutants including heavy metal, nutrient, pesticide and others. In a word, soil erosion degrades the soil resources that human sustain on and threatens the environment that people living in. Soil erosion model, which predicts total amount of eroded soil, helps to understand the erosion process and distribution, thus to guide the installations of soil conservation measures, received great attention both from the scientific field and from the governmental sectors. The soil erosion model can be generally divided into two categories, the empirical model and the physical-based model. The former is derived through the extensive analysis of numerous experimental data while the latter is developed upon the concrete physical processes, which confines the applications of empirical models to the areas that they derived from and give ways to physical-based ones for modeling in a wide range of areas. The physical-based models are usually distributed to account for the spatial variations of rainfalls and ground features. The development of distributed models was greatly pushed in the later 1990s as the advance of spatial information technologies. Remote sensing (RS) was believed to be a good data source to acquire ground cover information and digital elevation model (DEM) was proved to be useful data to derive topographical parameters of the watershed. Geographic Information Systems (GIS) have changed our way to handle regionally distributed information.China is one of the countries that were seriously suffered from soil erosion. In China, about 1,790,000 km2 land is suffered from water erosion, which accounts for 18.3% of China’s total area. As a result, 5 billion tons of soil was eroded and over 667 km2 arable land was lost each year. However, the soil erosion modeling techniques in China is not so advanced for the lack of good models and the difficulty in modeling over the complex topography in the country. The empirical models are frequently employed in soil erosion modeling since there were only a few distributed physical-based models developed. And they were mainly focused on the loess plateau areas. There are no distributed physical-based models available in Southern China, where soil erosion by water is also very seriously. So it’s urgent to develop physical-based soil erosion model that’s suitable for the Southern China area, esp. the Poyang Lake watershed.After a throughout review of physical-based models, this paper proposed a physical-based distributed soil erosion model to calculate total soil erosion amount and distribution after a single rain event. The calibration of the model was conducted using the data collected from the gauge stations. And the calibrated model was used to model the soil erosion after a storm and gave satisfactory results. This paper mainly include the following sections:1. The extraction of watershed parameters based on DEM. The extraction is consisting of the following steps. 1) The treatment of DEM to remove the sinks or flat area, 2) the assignment of flow direction of the DEM grids based on D8 algorithm, 3) the calculation of flow accumulation grids based on flow direction grids. The slope cells and channel cells were separated by setting a threshold to the flow accumulation grids. And the watershed boundary was identified by iterative search from the watershed outlet based on flow direction grids. This paper reviewed the DEM reprocessing methods and presented an effective algorithm based on three-direction search. Comparisons were made between the ArcHydro method and the proposed method. The results showed that the proposed algorithm can eliminate the parallel rivers as frequently occurred with ArcHydro method, and the algorithm made fewer elevation modifications to the original DEM.2. The study of the distributed hydrological module. Since soil erosion by water is closely related to rainfall and runoff, erosion modeling can’t be separated from the procedures used to model the generation of runoff and its routing down a hillside and through the river channel network. In this paper, saturation excess runoff was modeled after satisfaction of rainfall interception by vegetation and soil infiltration. The runoff was routed from the watershed to the outlet using the ranked-grid based routing method. Kinematic wave and diffusion wave were employed in the flow routing according to the slopes of the cells.3. The modeling of soil erosion processes. The basis for describing soil erosion processes in Southern China is rather limited. In this paper, we borrowed the equations derived overseas and in the loess plateau areas. Three main erosion processes such as soil detachment by raindrop impact, soil detachment by rill and inter rill flows were considered and modeled accordingly. Flow transport capacity was also modeled to calculate the total soil erosion yield in each cell. The net sediment yield in each cell was then routed downstream to the outlet using continuity equation.4. The calibration of the model. The model can’t be applied to study area before it was properly calibrated because some of the model parameters are empirically based. There are several calibration methods available. However, taking the massive data volume of distributed model into consideration, human-machine interactive method was selected to calibrate the model along with the rainfall and hydrological and sediment data collected from the gauge stations.5. The validation and application of the model. ZaoKouShui watershed, located in the upstream of Xiu River in the Poyang Lake watershed in Southern China was chosen as the study area because it’s one of the watershed that is suffered from severe soil erosion and it’s possible to collect the detailed rainfall and hydrological and sediment data of the watershed. The model was properly validated using the measured data. The statistics showed that the model was capable of modeling water and sediment yield at watershed scale. The modeling of soil erosion was conducted after one storm and gave the distribution of the erosion and total sediment yield at the outlet. The results were satisfactory compared with the measured data. And it was also revealed that the main sediment comes from the agriculture land cultivated on slopes higher than 15%.更多还原