【作者】 花利忠；

【导师】 贺秀斌；

【作者基本信息】 中国科学院研究生院（成都山地灾害与环境研究所） ， 自然地理学， 2007， 博士

【摘要】 水土资源是人类最宝贵的自然资源和生存基础,土壤侵蚀是导致土地资源退化乃至彻底破坏的主要原因。三峡库区的水土保持生态环境状况,不仅关系到三峡工程的安危和经济运行年限,而且还关系到长江中上游乃至整个长江流域经济的可持续发展。库区水土流失非常严重,已成为长江中上游的重要环境问题,但是该区研究基础薄弱、手段有限。当前在三峡库区的中、大流域尺度上,采用物理过程模型对侵蚀产沙时空变化特征的研究几乎是一片空白。本文以长江三峡库区腹心——大宁河流域为研究对象,在3S技术支持下构建了AnnAGNPS (Annualized AGricultural Non-Point Source)模型数据库,在模型校正与验证的基础上,利用模型研究了流域侵蚀产沙时空变化特征,分析了流域及其支流的SDR值,并模拟两种BPMs对流域泥沙量的影响,为控制三峡库区水土流失和实现流域优化管理提供重要科学依据。主要研究结论如下:一、深入分析流域气象、地形、土壤、土地利用和作物管理等相关参数,在3S技术平台上构建了适用于库区中等流域尺度侵蚀产沙的模型数据库。二、用数字滤波法把水文站径流中的基流分割后,利用偏差、决定系数和Nash-Sutcliffe模拟系数3个指标评价了模型的适用性,结果表明,模型能够比较理想的预测流域年和月水平的径流量和泥沙量,适用于该流域侵蚀产沙定量评价。(1)模型校正结果显示:年水平上,预测与实测径流深偏差为-1.09%;月水平上,两者最佳拟合直线斜率为0.94,相关系数r2为0.97,NS系数0.97。(2)模型径流验证结果显示:年水平上,预测与实测径流深偏差都在-12%以内;月水平上,相关系数r2都大于0.90,NS系数都超过0.80;3年整体比较显示,两者最佳拟合直线斜率为1.04,相关系数为0.93,NS系数为0.92。(3)模型泥沙量验证结果显示:预测与实测的多年平均输沙量偏差为1.8%。三、探明了流域侵蚀产沙时空变化规律:(1)流域无明显流失面积1 996.8 km2,占流域总面积的47.76%,流域平均土壤侵蚀模数为1 551.7 t·km-2·a-1,流域流失面积(不含无明显流失面积)的侵蚀模数为2 650.6 t·km-2·a-1;(2)流域土壤流失量和输沙量空间差异较大,流域年均土壤侵蚀总量为648.8万t·a-1,其中136.6万t·a-1 (21.05%)泥沙沉积于流域各分室内,317.6万t·a-1 (48.97%)泥沙沉积于流域沟道系统中,其余194.6万t·a-1 (29.98%)泥沙直接输入长江干流。(3)流域内不同土地利用类型侵蚀量分析显示:坡耕地和低覆盖度草地是流域水土流失的策源地,占流域面积22.33%的耕地和5.34%的低覆盖草地侵蚀量分别占流域侵蚀总量的55.59%和10.90%。(4)侵蚀产沙季节分布集中程度很高,呈典型的单峰型。6-8月是产流产沙集中期,分别占年总量的64.26%和74.26%,其中7月份为产流产沙最高峰,分别占年总量的27.58%和34.44%。(5)侵蚀产沙年季分布差异大,径流和泥沙变异系数分别为0.45和0.41,径流深和泥沙量年季变化趋势成正相关(y = 2.30x, r2 = 0.83);产沙量不仅与年径流总量有关,而且与产流分布和强度直接相关。四、揭示了流域泥沙输移特性:(1)大宁河干流多年平均输沙量自上游巫溪站的124.6万t,沿程递增至大昌站的178万t,大昌站以下增加缓慢,到流域出口为194.6万t。6条支流中,西溪河对流域出口的径流量贡献最大,占20.97%。对流域出口泥沙量贡献最大是东溪河,占26.88%;巫溪站以下至大昌站的大宁河干流左岸区间以坡耕地和低覆盖度草地为主,而且年均径流深大,也是产沙集中区,如柏杨河输沙量占流域出口16.24%,年均输沙模数高达1 111.2 t·km-2·a-1。(2)上游(巫溪)——中游(大昌)——流域出口的3个干流站点控制面积其输沙模数(623.1 -594.1- 465.3 t·km-2·a-1)与SDR (0.440- 0.331-0.300)逐渐减小,两者呈正相关关系,流域面积与SDR呈负相关关系。但从6条支流看,这种关系不明显,SDR值最高的是面积最大的西溪河0.531,SDR值最低的是后溪河0.376,输沙模数最高的是面积最小的柏杨河1111.2 t·km-2·a-1,输沙模数最低是平定河402.8 t·km-2·a-1。(3)大宁河流域的SDR值为0.30,与相关文献的结果比较接近。五、评价了两种最佳管理措施(BMPs)的水土保持效益,结果显示,退耕还林是一种较好的治理措施,所有大于25°坡耕地种植灌木林后,将削减流域总侵蚀量和输沙量的24.5%左右。更多还原

【Abstract】 Water and soil resources are most important for human beings survival and development. Soil erosion is a major cause for land resource degradation and the total destruction.The Three Gorges Reservoir (TGR) region is a very severely eroded region in China. The soil erosion and sedimentation of this region have been concerned by government and researchers for years. But the research techniques in this region are traditional and inefficient. The study on tempo-spatial dynamic changes of soil erosion and sediment using physically based distributed model at the large-watershed scale is almost a bank research branch in this region.AnnAGNPS, a continuous-simulation, watershed-scale model, is widely used to evaluate non-point source pollution from agricultural watersheds in a number of countries. AnnAGNPS simulates surface water, sediment, nutrients, and pesticides leaving the cells and their transport through the watershed. It is an expansion of the capabilities of the single event AGNPS. The watershed is subdivided into homogenous land areas (cells) with respect to soil type, land use, and land management. These areas can be of any shape from the original square grid cells of AGNPS to more appropriate hydrologic boundaries that can be generated by terrain-following Geographical Information System (GIS) software. The model enables managers to identify sensitive or critical areas of non point source pollution and to perform and evaluate various“what-if”scenarios in the decision-making process related to watershed management and specific water-quality management.The Daning River Watershed with a contributing drainage area of about 4 181 km2 was selected as the case study. A large number of basic data including digital elevation model (DEM), landuse, soil category and crop management, were collected through the extensive on-site investigation to set up model databe. Based on the calibration and validation of the model, the model was applied to evaluate the tempo-spatial dynamic changes of erosion and sediment, analyze the sediment delivery ration (SDR) of Daning River Watershed and its six tributaries and simulate two best management practices. The main conclusions are drawn as follows:With the aid of 3S (GIS, Remote Sensing, Global Position System) technology, the AnnAGNPS model database was set up in the area by analysis of rainfall characteristics, size of simulating cells, translation of soil relative parameters, characteristic of land use management, and so on.The model was calibrated using the monitoring runoff data at Wuxi hydrological station in 1998. The model was validated on the basis of the monitoring runoff data between 2003 and 2005 at the station and average annual sediment loading data from Dachang hydrological station. The calibration result for runoff is very satisfactory (relative errors -1.09% by year and Nash-Sutcliffe coefficient (NS) 0.97 by month). The regression of the monthly predicted runoff with the observed runoff on the line of equal values is good with a R-square value of 0.97 and a slope of 0.94. The monthly predicted runoff also matches well with the observed runoff in the validation (relative errors < -12% by year and r2> 0.90, NS > 0.80 by month). The model predicted average sediment loading over the eight-year period from 1998 to 2005 is more than 1.8% of the observed data. The predicted runoff and sediment loading compares fairly well with the observed data indicating that the model has an acceptable performance in simulation of runoff and sediment in Daning River Watershed.The tempo-spatial variation of soil erosion and sediment of the watershed is found.The tiny erosion area of the Daning River Watershed has come to 1 996.8 km2, accounting for 47.76% of the total watershed area. The annual average soil erosion rate of the watershed is found to be 1 551.7 t·km-2·a-1, and the rate is 2650.6 t·km-2·a-1 if tiny erosion area is not included. The soil erosion rate and erosion amount vary greatly among different land use types. The sloping lands and grasslands with low coverage cover 22.33% and 5.34% of the total area respectively, however, their erosion amounts accounted for 55.59% and 10.90% of that in the watershed. The study demonstrates that sloping lands and grasslands with low coverage is the main source of soil loss and the key of soil loss control is to utilize them rationally.The sediment budget of the watershed clearly shows that the average annual erosion amount of the Daning River Watershed is 6.488 million tons, 48.97% of the sediment is deposited in the reach networks, 21.05% is deposited in the cells and the rest is the output of the watershed. In the Daning River Watershed, runoff is positively correlated with sediment loading and annual and interannual variations of them vary greatly. The runoff and sediment losses mainly occurred between June and August. More than 64% and 74% of total runoff and sediment loading, respectively, are lost during this period. July is the peak month and more than 27 % and 34% of total runoff and sediment loadings are lost. Sediment loading is related to total annual runoff, especially to runoff distribution and intensity.The characteristics of sediment delivery are studied. Xixi River, one of the biggest tributaries in the Daning River Watershed, has largest runoff loading contributions up to 20.97% of the watershed outlet. But Dongxi River, the second tuiburary, has largest sediment loading contributions up to 26.88% of the watershed outlet. The area of left bank from the watershed main stream from Wuxi to Dachang is also the sediment source. Viewed from three main stream stations, i.e. Wuxi, Dachang and the watershed outlet, the sediment delivery ratio (SDR) is positively correlated with sediment loading and negatively correlated with the watershed area. But the relations are not significant. In the six tributaries, the maximum SDR is Xixi River up to 0.531 and the minimum SDR is Houxi River up to 0.376. The SDR of Daning River Watershed is 0.30, consistent with relevant conferences.Two BMPs have been made quantitative assessment with the AnnAGNPS before and after applied, and this method gives the scientific basis for soil erosion and sediment controlling measures. Reconverting farmland with larger than 25 gradients to forest with medium coverage will reduce 24.5% of the total erosion amount and sediment loading.The study demonstrates that AnnAGNPS model is successfully applied to research and control soil and water loss of Daning River Watershed, Three Gorges Reservoir Region. The model is a research tool and practical means for optimal watershed management and rational landuse resources, and can be applied for other comparable areas of the Three Gorges Reservoir Region.更多还原