中文摘要：

地形因子是影响土壤侵蚀的重要因素，包括坡度、坡长等，通常基于DEM来提取，但目前构建的DEM总是局限于反映连续光滑的自然坡面，无法表现出梯田等人工地形的信息。在黄土高原地区，通过多年的治理，修筑了大面积梯田，极大地改变了地表微形态，影响了坡度和坡长，进而影响土壤侵蚀定量分析。以安塞县的梯田为实验对象，采用基于真实田坎的方法，构建嵌入梯田信息的DEM（梯田DEM），并对原始DEM与梯田DEM提取的坡度、坡长等地形因子以及计算得到的坡度坡长因子进行对比分析。结果表明：（1）基于真实田坎方法构建的梯田DEM，能较好地反映田坎位置及形态；(2)与样区原始DEM相比，加入梯田信息后，地形因子的栅格数大幅度增加且基于梯田DEM提取的坡度、坡长、LS因子明显减小，对土壤侵蚀定量分析有一定影响。该研究对地形因子提取和土壤侵蚀评价具有重要意义。

英文摘要：

Terrain factors include slope, slope length, etc. They are primary elements of soil erosion, which can impact surface runoff, erosion, sediment delivery, the influx and transport of aqueous and sand material. As a function of slope and slope length, LS factor is able to influence the appraisal of regional soil erosion as well. Interpolating the elevation information based on existing topographic map is the most common method for constructing DEM. Building terraces is an efficient facility for soil and water conservation engineering. However, they cannot be presented appropriately in topographic maps. Therefore, current DEMs are limited to represent natural landforms. They are unable to show the information of artificial landform like terraces. Moreover, the terrain factors based on DEM can not reflect the influence of terraces. Actually,large areas of terraces have been built in the Loess Plateau after several years’ management. The terraces affect surface micro-morphology drastically, and make a great influence on the slope, slope length and quantitative analysis of soil erosion.We choose terraces in a basin in Ansai County as our case study area. The involved materials include DEM of Ansai(its scale is 1:10000, resolution is 5m) and a remotely sensed image with a resolution of 2.5m. Based on the improved rapid modeling method for terrace digital elevation model, we adopt the actual ridge-based method to construct the DEM with terraces’information (T-DEM), by integrating with high resolution remote sensing image. Using Spatial Analyst Module of ArcGIS and LS TOOL, slope, slope length and LS factor were extracted from original DEM and T-DEM, respectively. In comparison with the original DEM in sample region, slope, slope length and LS factor extracted by T-DEM are smaller and the quantity of raster cells is higher. More precisely, the average of slope decreased by 1/4 and its range increased. Slope length reduced by 5/6 and the proportion of short slope increased a lot. The positive correlation between slope,