【作者】 何建林；

【导师】 何丙辉；

【作者基本信息】 西南大学 ， 水土保持与荒漠化防治， 2010， 硕士

【摘要】 土壤侵蚀,是导致土地资源退化和损失的主要原因,是限制当今人类生存与发展的全球性环境灾害。土壤侵蚀预报模型用于了解土壤侵蚀过程与强度,是进行土壤侵蚀动态监测预报的核心工具和难点问题,体现着土壤侵蚀定量化研究的成果。WEPP模型是迄今为止较为复杂的基于连续事件的分布式参数土壤侵蚀模型,其需要考虑气候、地形、土壤和管理四大方面的影响因素,具有外延性好、反映侵蚀分布好、模拟侵蚀物质输移过程好的优点。已有对WEPP模型在紫色土区的研究主要是部分预测参数及其在休闲地上的研究,因此本论文利用四川盆地中部丘陵区遂宁水土保持试验站2006～2008年四个标准径流小区(10°农耕地小区、10°果园地小区、15°农耕地小区和15°林地小区)的29次单次降雨侵蚀实际观测资料,通过相对误差、模型有效性系数(ME)、Pearson相关系数、敏感度值(S)、变化倍数等指标,对WEPP(坡面版)模型预测参数在紫色土丘陵区的建立、验证和敏感性进行研究。本研究共分三个部分,第一部分是WEPP模型预测参数的建立,采用单一直线坡建立坡度为17.63%(10°)或26.79%(15°),坡长水平投影为20m,坡侧面宽度为5m的地形参数；选择CLIGEN Generated的方式,利用遂宁站内气象观测场1985～2008年降雨和气温观测数据及美国得克萨斯州(TEXAS)森特维尔(CENTERVILLE)参证站点数据建立气象参数；直接选取WEPP模型中自带的措施文件建立管理参数；在土样采集、土壤理化性质测定的基础上,采用计算输入或模型生成的方式建立土壤参数。第二部分是不同预测参数的验证,通过相对误差、模型有效性系数、Pearson相关系数等指标,以29次实测单次降雨侵蚀产流量和产沙量数据与不同预测参数组合下WEPP模型模拟预测得到的对应数据进行比较,研究表明：(1)计算输入好于模型生成的土壤参数建立方式,且产流量预测效果好于产沙量的预测。手工计算对产流量的相对误差为±5%、±10%、-5%～-50%和-30%～-100%：模型有效性系数ME为0.982、0.990、0.789、-2.645和Pearson相关系数为0.994、0.996、0.970、0.365,按其排序为10°果园地小区>10°农耕地小区>15°农耕地小区>15°林地小区。当用产沙量进行验证时,10°和15°农耕地小区的手工计算方式模拟相对误差(-60%～-100%%和-80%～-100%)略好于模型计算方式(-75%～-100%%和-85%～-100%),10°果园地小区和15°林地小区的相对误差一样(-65%～-100%%和-100%)：10°和15°农耕地小区手工计算方式模拟的ME值为-0.555和-1.035；手工计算方式模拟的Pearson相关系数排序为10°果园地小区(0.866)>15°农耕地小区(0.612)>10°农耕地小区(0.378)。(2)对10°和15°农耕地小区的“玉米+红苕+小麦”种植模式所选取模型自带的“Com,soybean,wheat, alfalfa(4yrs)-consv till.rot、Corn, soybean, wheat, alfalfa(4yrs)-conv till.rot和Corn, soybean,wheat, alfalfa(4yrs)-no till.rot"三种措施文件得到相同的产流量或产沙量,其对应的ME值为0.982或-0.555和0.789或-1.035,Pearson相关系数为0.994或0.866和0.970或0.612；对10。果园地小区选用“30%Cover after fire. rot"措施文件时,其预测产流量和产沙量的ME值为0.990和-0.758,Pearson相关系数为0.996和0.866；对15°林地小区的五种模型自动措施文件都不理想,但"Tree-5 yr old forest.rot"的ME值(-2.645)略好于其他四个措施文件(-3.540)。第三部分是不同预测参数的敏感性分析,通过相对误差、变化倍数和敏感度值等指标,以10°或15°农耕地小区和10°果园地小区中具有代表性的2008年8月24日单次降雨侵蚀的产流量或产沙量为基准值,15°林地小区选用2007年6月28日,进行单参数在±20%、±40%、±60%、±80%、+100%范围内变化的敏感性分析,结果表明：(1)坡度对产流量敏感,且为正相关,按S值排序为15。林地小区(3.993)>10°农耕地小区(0.133)>10°果园地小区(0.132)>15°农耕地小区(0.063)；10°或15°农耕地小区对产沙量不敏感(S值为0.000),仅10。果园小区对产沙量中度敏感(S值为0.531),且为正相关。(2)降雨量、降雨历时、最大雨强和TP(%)的四个指标对10°农耕地小区、10°果园地小区、15°农耕地小区的产流量和产沙量达中度或高度敏感,除降雨历时为开口向上先减小再增大的二次曲线外,其他三个指标均为正相关,其变化程度为降雨量>最大雨强>最大历时>TP(%)；降雨量对产流量和产沙量的敏感度值排序为15°农耕地小区(2.262和2.394)>10°农耕地小区(2.201和2.255)>10°果园地小区(2.193和1.704)；在15°林地小区中最大雨强(S值为0.027)和TP(S值为0.000)对的产流量不敏感,但降雨量(S值为2.604)和降雨历时(S值为2.283)高度敏感,且除基准值外,无法模拟到产流量预测值。(3)土壤反照率、细沟侵蚀值、临界剪切力在10°农耕地小区、10°果园地小区、15°农耕地小区对产流量或产沙量都不敏感(S值均为0.000)；细沟间侵蚀值对四个小区的产流量都不敏感(S值都为0.000)；15°林地小区的土壤反照率和细沟侵蚀值对产流量也不敏感(S值为0.000),而临界剪切力却高度敏感(S值为1.152)；初始饱和度和有效水力传导率对产流量和产沙量都高度敏感,初始饱和度对产流量正相关,有效水力传导率对产流量或产沙量都是负相关；细沟间侵蚀值对产沙量正相关,其S值大小为10°农耕地小区(1.281)>10°果园地小区(0.444)>15°农耕地小区(0．385)。更多还原

【Abstract】 Soil erosion, as the main reason of leading degradation and loss of land, is the global environmental disaster which limits human existence and development. Soil erosion prediction model is applied to find out the processes and intensity of soil erosion and is the core tool used to monitor and predict trends of soil erosion, which embodies achievement of the quantitative study of soil erosion. WEPP model is comparatively complex soil erosion model of distributed parameter based on continuous events. It entails consideration of the four influencing factors, that is, climate, landform, soil and management. The model is with good extension, and can reflect erosion well and the transferring process of simulating erosion materials is good. Previous studies of WEPP model have been mainly on the prediction parameters and their corresponding researches in the fallow lands. Therefore, through such indexes as relative error, coefficient of efficiency of the model (ME), relative Pearson coefficient, sensitivity value(S) and variation multiples etc, and this paper systemically and integrally does some research on the establishment, validation and sensitivity of prediction parameter of WEPP model (slope version) in purple soil areas based on actual observation data of 29 times of rainfall erosion from four standard runoff plots (farmland with 10°, orchard land with 10°, farmland with 15°and woodland with 15°) of Suining Water Conservation Experiment Station in hilly areas of Sichuan Basin from 2006 to 2008.This study is divided into three parts. The first part is the establishment of the WEPP model prediction parameters. Based on single linear slope, a topographic parameter is established with a gradient of 17.63%(10°), or 26.79%(15°), horizontal projection of slope length as 20m, width of the side slope as 5m; way of CLIGEN Generated is chosen; The meteorological parameter is established by using observation data of rainfall and temperature in Suining Station from 1985 to 2008 and data of Centerville Reference Station of Texas in the United States:management parameter is established by select the measure files in the WEPP model directly; based on soil sample collection and measured data of physical and chemical properties of soil. soil parameter is built by manual calculation or model input generationThe second part is the verification of different prediction parameters. Through the relative error, model efficiency coefficient, relative Pearson coefficients as well as comparison of relevant data obtained by simulating prediction of WEPP model under the combination of statistics of measured erosion and sediment production of 29 times of rainfall with different prediction parameters, study shows that:(1) Manual calculation is better than the establishing method of soil parameter generating from model calculation. Besides, effect of runoff prediction is better than prediction of sediment production. The relative error of runoff by manual calculation are=5%.±10%,-5%～-50%.-30% and-100%. Model efficiency coefficients (ME) are 0.982.0.990.0.789 and-2.645. Pearson coefficients are 0.994,0.996,0.970, and 0.365. According to the two coefficients above the order is like this:orchard land with 10°> farmland with 10°>farmland with 15°> woodland with 15°. When the sediment production is used for validation. simulating relative error of manual calculation in farmlands with 10°nd 15°(-60%-100%% and-80%-100%) is slightly better than model calculation (-75%-100%% and-85%～-100%). Relative error in orchard land with 10°is the same with woodland with 15°(-65%-100% and-100%). Simulating ME value of manual calculation in farmland with 10°and 15°are-0.555 and-1.035. The order of Pearson coefficient of simulation by manual calculation is orchard land with 10°(0.866)> farmland with 15°(0.612)>farmland with 10°(0.378). (2) The same runoff and sediment production art obtained through the three kinds of measures. namely, corn, soybean, wheat, alfalfa(4yrs)-consv till.rot:corn, soybean. wheat. alfalfa(4yrs)-conv till.rot; and corn, soybean, wheat, alfalfa(4yrs)-no till.rot. in the" sweet potato-corn+wheat"planting patter in farmland with 10°and 15°. The corresponding value of ME is 0.982 or-0.555. and 0.789 or-1.035. The Pearson coefficient is 0.994. or 0.866.0.970 and 0.612. As the measure file "Cover after fire. rot" is applied to orchard land with 10°, ME values of predicted runoff and sediment production are 0.990 and-0.758. The Pearson coefficients are 0.996 and 0.866. Automatic measures of the five models are not appropriate to woodland with 15°. Whereas ME value of the "Tree-5 yr old forest.rot" (-2.645) is much better than the other four measure files (-3.540).The third part is the sensitivity analysis of different prediction parameters. Through such indexes as relative error, variation multiple and sensitivity value etc, the runoff and sediment production of single rainfall erosion on 28th June,2008 in farmland with 10°or 15°and orchard land with 10°re used as reference values. Woodland with 15°on 28th June.2007 is selected. Sensitivity analysis with single index which change within±20%,±40%,±60%,±80%.and+100% is made. Results show that:(1) Slope is sensitive to runoff production and there is positive correlation between them. According to the value of S, the order is woodland with 15°(3.993)>farmland with 10°(0.133)> orchard land with 10°(0.132)> farmland with 15°(0.063). Farmland with 10°or 15°is not sensitive to runoff production(S=0.000). Orchard land with 10°is moderately sensitive to runoff production (S=0.531) and the correlation is positive.(2) The four indicators of diachrony of rainfall, rainfall, maximum rainfall intensity and TP (%) are moderately or highly sensitive to the runoff and sediment production in farmland with 10°, orchard land with 10°, and farmland with15°. The rainfall diachrony is a conic opening up which first decreases and then increases. The other three indicators are positively related to the runoff and sediment production. The degree of change is rainfall>maximum rainfall intensity> maximum diachrony>TP (%). The order of sensitivity value of rainfall to runoff and sediment production is farmland with 15°(2.262 and 2.394)> farmland with 10°(2.201 and 2.255)> orchard land with 10°(2.193 and 1.704); The maximum rainfall intensity(S=0.027) and TP (S=0.000) in woodland with 15°re not sensitive to runoff production but the rainfall (S=2.604) and rainfall diachrony (S=2.283) are highly sensitive to the runoff production. The predicted runoff production can not be simulated except the reference value.(3) Soil albedo, the value of rill erosion, and the critical cut force are not sensitive to runoff and sediment production in farmland with 10°, orchard land with 10°and farmland with 15°(S=0.000). The inner rill erosion value is not sensitive to the runoff production in the four areas either (S=0.000), while the critical cut force is highly sensitive to the runoff (S=1.152). Initial saturation and the effective hydraulic conductivity are both highly sensitive to runoff and sediment production and there is positive correlation between initial saturation and runoff production. The effective hydraulic conductivity is negatively correlated with runoff or sediment production. The inner-rill erosion value is positively related to sediment production. The order of S is farmland with 10°(1.2813)> orchard land with 10°(0.444)> farmland with 15°(0.385).更多还原