【作者】 王春梅；

【导师】 李锐； 杨勤科；

【作者基本信息】 中国科学院研究生院(教育部水土保持与生态环境研究中心) ， 土壤学， 2012， 博士

【摘要】 坡度是影响土壤侵蚀的主要地形因子之一。在区域尺度土壤侵蚀研究中，坡度参数一般是通过中低分辨率DEM来提取的。但随着DEM分辨率的降低，其上提取的坡度发生衰减，使之不能有效反映地形与土壤侵蚀的关系，从而降低区域土壤侵蚀模拟精度。本研究针对以上问题，将数字图像处理、数字地形分析和地统计理论有机结合，通过对坡度与分辨率关系的系统分析，认识坡度衰减规律，解释坡度衰减机理，提出坡度理论分布模型，并在此基础上结合直方图匹配原理，推导得出坡度降尺度变换数学模型，在全国范围内给出坡度降尺度变换（将基于90m分辨率SRTM数据计算的坡度数据降尺度变换使其与25m分辨率坡度有相似的统计特征）数学模型关键参数，最后以区域土壤侵蚀定性评价为例，对坡度降尺度变换的实用性做出评估。本论文在全国范围内选取60个样区，其中有6个分析样区，用于探索坡度衰减规律和机理，验证坡度理论分布模型和坡度降尺度变换数学模型；有53个验证样区，用于验证全国坡度降尺度变换数学模型关键参数的合理性；有1个土壤侵蚀评价样区，用于对比坡度降尺度变换前后土壤侵蚀评价结果的差异性。论文主要研究结果如下：（1）坡度基本统计特征及其随分辨率的变化情况在不同地形类型区存在差异。平均坡度和坡度标准差随分辨率降低而减小，但在不同地形类型区随分辨率变化的程度不同。黄土丘陵沟壑区（LP样区）和南方红壤丘陵区（SE样区）变化最大；东北漫川漫岗丘陵区（NE样区）和陕北风沙过渡区（NS样区）变化最小；华北土石山区（SD样区）变化中等；四川紫色土丘陵区（SC样区）在较高分辨率范围内变化较小，在较低分辨率范围内变化增大。随分辨率降低坡度频率曲线和累积频率曲线向低坡度范围移动，不同地形类型区移动的程度不同。LP样区和SE样区频率曲线随分辨率变化最大，NE样区和NS样区变化最小，SD样区变化中等，SC样区在较高分辨率范围内变化较小，在较低分辨率范围内变化增大。（2）通过独立结构变异函数理论模型（Independent Structures Model, ISM）拟合得出坡度各空间频率组分随分辨率降低的变化特征以及不同地形类型区各组分的差异性。随分辨率降低坡度整体变异性减小，相关距离增大；随分辨率降低受影响最大的为空间高频组分，而空间低频组分随分辨率变化较小；在6个分析样区的模拟结果表明，NE样区和NS样区坡度信息在空间低频组分占的比重较大；LP样区和SE样区在空间高频组分占的比重较大；SC样区在空间中频组分占的比重较大；SD样区在各个空间频率组分所占比重居中。以上结论较好地解释了分辨率降低过程中不同地形类型区坡度变化的差异性。（3）将高程梯度分析和卡方分布理论结合，完成了坡度理论分布模型的推导，其主要参数为高程梯度p和q的标准差。模型精度取决于样区内部地形是否单一，高程梯度p和q是否符合模型前提假设（即p和q呈正态分布，相互独立，均值为0，方差相等）。将坡度理论分布模型在6个分析样区系列分辨率上进行应用及验证，结果表明本研究提出的坡度理论分布模型可以模拟坡度频率曲线、累积频率曲线、平均值和标准差，在6个分区样区上模型精度较高。（4）结合坡度理论分布模型与直方图匹配原理推导得出坡度降尺度变换数学模型，其最重要的参数为高低分辨率高程梯度p和q的标准差。降尺度变换数学模型在6个分析样区模拟效果较好；基于全国25m分辨率Hc-DEM得到全国范围内坡度降尺度变换重要参数（高分辨率高程梯度p和q的标准差）的参考值并在53个验证样区上通过了验证。（5）以土壤侵蚀分类分级标准（SL190-2007）所规定的水蚀面蚀评价方法为土壤侵蚀强度评价依据，以安塞县为研究区，基于90m SRTM计算的坡度（以下简称SRTM坡度）、降尺度变换后90m SRTM坡度（以下简称降尺度变换坡度）和25m分辨率坡度分别进行土壤侵蚀强度评价，探讨坡度降尺度变换对区域土壤侵蚀评价的影响，并对比基于三种坡度数据的土壤侵蚀强度评价结果。从图面效果和统计信息来看，基于降尺度变换坡度和基于25m分辨率坡度进行的土壤侵蚀强度评价结果相似，而基于SRTM坡度的土壤侵蚀强度评价结果与前两者有较大差别，其土壤侵蚀强度评价结果偏弱。坡度降尺度变换对区域土壤侵蚀评价是必要的。更多还原

【Abstract】 Slope gradient is one of the main factors in soil erosion. In the research of regional soilerosion accessment, slope gradient is generally calculated from lower resolution DEM.However, slope gradient becomes smaller with resolution becomes coarser, thus makes itdifficult to show the relationship between slope gradient and soil erosion and makes themodeling result of regional soil erosion poor. In this research, the digital image process,digital terrain analysis and geostatistical analysis were mixed together. We studied therelationship between slope gradient and resolution, explained the reason of slope reductionwith resolution, build a theoretical slope distribution model, got a mathmetial slopedown-scaling model and the key parameter in the model over China(downscale from90mSRTM to25m) and evaluated the change of regional soil erosion accessment result because ofslope down-scaling.There are60research sites in this paper. Six of them are called analyse-samples which aredused in study of slope reduction with resolution, validation of theoretical slope distributionmodel and mathmetial slope down-scaling model. Fifty three of them are called downscalingvalidation-samples which are used in validation of parameters of mathmetial slopedown-scaling model over China. One of them are called soil erosion accessment-sampleswhich is used in comparation of soil erosion accessment based on coarser and down-scalingslope. The results of this research are as follows.(1)There are some differences between varied terrain types in the change of basic slopestatistics with resolution. Mean slope and slope standard deviation (STD) reduce withresolution becomes coarser. However in different terrin types the change is different. In LoessHilly area (LP sample) and Red Soil Hilly areas (SE sample) the change is most intensive. InRocky Area of Shandong province (SD sample) the change is moderate. In Northeast GentleHilly Black Soil area (NE sample) and Wind Drift Sand area in North Shannxi Province (NSsample) the change is most gentle. In Purple Soil Hilly area in Sichuan province the change is gentle when resolution is finer and is more intensive when resolution is coarser. Frequencyand Cumulate Frequency curves moves to gentle slope with resolution become coarser. Thechange in different terrain types is similar as the change of mean slope and STD of slope.(2) Semi-variogram is a useful tool in analyse of terrain structures. Change of slopeinformation in different scale components with resolution and terrain types can be detected.The overall variance of slope reduces and correlation distance becomes larger with resolutionbecomes coaser. The result of ISM model shows that with resolution becomes coarser thechange of high frequency component of slope is most significant and that of low frequencycomponent is small. In NE and NS sample low frequency component of slope is moreimportant than in other samples. However in LP and SE sample high frequency component ofslope is more important than in other samples. In SC sample moderate frequency componentof slope is more important than in other samples. In SD sample the proportion of high, lowand moderate frequency component in overall variance is moderate in the six samples. Theresult well explained the reason for the difference in slope scale effect with resolution indifferent terrain types.(3) This paper got a theoretical slope distribution model based on slope gradient of DEMsurface and The chi-square distribution. The main parameter in this model is std of slopegradient p and q. The model efficiency is dependent on whether the terrain type is single inthe study area that is whether p and q is normally and independent with zero mean and eaqualvariance. The model has been used in the six analyse-samples at series of resolutions. Theresult shows that the model used in this paper can model slope frequency and cumulatefrequency curves, mean and std of slope. The model efficiency is high in the sixanalyse-samples.(4)Slope down-scaling mathematical model was got based on theoretical slope distributionmodel and histogram matching method and applied in the six analyse-samples. The mostimportant parameter in the slope down-scaling mathematical model in this paper is the std ofp and q at high resolution. The paper offers the std of p and q at resolution of25m based on25m Hc-DEM all over China and the parameter was validated at the53downscalingvalidation-samples.(5) In this paper the method showed in Standards for classification and gradation of soilerosion (SL-190-2007) was used as the method of soil erosion assessment. The study area isthe county of Ansai. In order to know the influence of slope-downing to soil erosionassessment, the slope based on90m SRTM (SRTM slope for short) was downscaled throughthe slope down-scaling mathematical model to25m. Then soil erosion grade was calculatedbased on SRTM slope,25m slope and the downscaled slope. The result shows that the erosion grade result is similar based on25m slope and the downscaled slope. However there isdifference between erosion grade result based on SRTM slope and the other two slopesurfaces, the erosion grade result based on SRTM is relatively gentle. The slope downscalingprocess in regional soil erosion assessment is essential.更多还原