水系路径、小流域边界、倾泄点位置等水文特征的高精度提取是进行水文特征分析的基础和前提。为了解决低分辨率SRTM DEM、ASTER GDEM等DEM（Digital Elevation Model）数据提取水文特征精度较低的问题，本研究首先基于InSAR（Interferometry Synthetic Aperture Radar）技术与Sentinel-1A的SLC影像数据得到空间分辨率与测高精相对较高的研究区DEM，然后对比分析基于不同DEM数据的小流域水文征提取结果，最终本研究主要得出以下结论：①基于SRTM DEM、ASTER GDEM及InSAR DEM的高程提取误差均值分别为8.1、8.5及3.7 m，其中InSAR DEM的高程提取精度最高；②基于SRTM DEM、ASTER GDEM及InSAR DEM提取6条干流水系路径的重合度分别为74.12%、85.50%及88.36%，重合度依次升高；③基于SRTM DEM、ASTER GDEM和InSAR DEM提取各水系倾泄点位置与实际倾泄点位置的均值误差分别为4.93、4.69及4.63 km，误差依次降低。

The high-precision extraction of hydrological characteristics， such as water system route， small watershed boundary and the position of pour point， is the basis and premise of accurate analysis of various hydrological characteristics. In order to solve the problem of low accuracy extraction results of hydrological feature based on low resolution DEM data （SRTM DEM， ASTER GDEM）， this study first obtains relatively high accuracy DEM with spatial resolution and altimetry precision based on InSAR technology and Sentinel-1A/SLC data， and then compares and analyzes the extraction results of hydrological characteristics for small watershed based on different DEM data sources. Finally， the main conclusions of this study are as follows： ① Based on SRTM DEM， ASTER GDEM and InSAR DEM， the extraction errors of mean height of five sampling points are 8.1， 8.5 and 3.7 m respectively， and the InSAR DEM has the highest accuracy. ② Based on SRTM DEM， ASTER GDEM and InSAR DEM， the coincidence degrees of the six main water system routes are 74.12%， 85.50% and 88.36% respectively， and the coincidence degrees increases successively. ③ The mean error between the actual position of pour point and the extracted position of pour point from SRTM DEM， ASTER GDEM and InSAR DEM are 4.93， 4.69 and 4.63 km respectively， and the error decreases successively.

Fine-resolution Light Detection and Ranging (LiDAR) data often exhibit excessive surface roughness that can hinder the characterization of topographic shape and the modeling of near-surface flow processes. Digital elevation model (DEM) smoothing methods, commonly low-pass filters, are sometimes applied to LiDAR data to subdue the roughness. These techniques can negatively impact the representation of topographic features, most notably drainage features, such as headwater streams. This paper presents the feature-preserving DEM smoothing (FPDEMS) method, which modifies surface normals to smooth the topographic surface in a similar manner to approaches that were originally designed for de-noising three-dimensional (3D) meshes. The FPDEMS method has been optimized for application with raster DEM data. The method was compared with several low-pass filters while using a 0.5-m resolution LiDAR DEM of an agricultural area in southwestern Ontario, Canada. The findings demonstrated that the technique was better at removing roughness, when compared with mean, median, and Gaussian filters, while also preserving sharp breaks-in-slope and retaining the topographic complexity at broader scales. Optimal smoothing occurred with kernel sizes of 11–21 grid cells, threshold angles of 10°–20°, and 3–15 elevation-update iterations. These parameter settings allowed for the effective reduction in roughness and DEM noise and the retention of terrace scarps, channel banks, gullies, and headwater streams.

从数字高程模型提取的汇水网络和汇水区等信息是分布式水文模型及应用分析的基础参数，基于地表汇水模拟的算法是提取该类信息的主要方法，其中，水流方向的确定对提取结果有着直接的影响。单流向算法因其易于实现、易于确定上游汇水区等特性，得到了广泛应用，然而单流向算法在坡度平缓区域会产生不自然的平行径流，能模拟地表水流分散径流特点的多流向算法可以在一定程度上避免此问题，但多流向算法使得不同区域的汇水单元可能存在交叉。本文结合两类流向算法各自的优点和适用性，设计实现了一种混合流向算法，以期在不同的地形条件下模拟得到更加合理的水流分配。首先，使用基于模板的形态检测方法，在给定阈值的基础上，对数字地形进行了分类，DEM被划分为山谷、山脊、鞍部、缓坡和陡坡5类。对陡坡、山谷和山脊区域运用单流向算法；对缓坡和鞍部区域采用多流向算法确定径流方向并进行水量分配。本文选取了黄土地貌和中低山丘陵的两个流域作为研究区，利用并采用了30 m和90 m两个分辨率的DEM。本文研究将混合流向算法与现有其他算法的结果进行比较。相比于多流向算法，该算法结果中的分散效应受到明显的抑制，相比于单流向算法，非自然的平行径流也大幅减少。同时，混合流向算法在较大分辨率DEM上（30 m）改进效果更加明显。

Hydrological information extracted from digital elevation models (DEMs) is the basis of distributed hydrological models. Algorithms determining water flow direction over terrain surface are basis for extracting hydrological information from the DEM. Flow direction and accumulation distributions have a direct effect on catchment area. Single flow direction algorithm is widely used because of its easiness to implement and to track upstream catchment areas. However, the single flow direction (SFD) algorithms tend to produce unnatural parallel flow paths in gentle slope areas. Therefore, multiple flow direction (MFD) algorithms have been proposed to calculate water flow directions. However, MFD algorithms create overlap upstream boundaries of watersheds. Considering applicability of SFD and MFD algorithms, a hybrid approach is proposed to calculate a more reasonable distribution of water under different terrain conditions. First, a template-based terrain classification algorithm is applied to the given DEM which is categorized into five classes:valleys, ridges, passes, gentle slope area and steep slope area with a slope threshold value. A single flow direction algorithm is applied to steep slope, valley and ridge area. A multiple flow direction algorithm is used to determine water flow directions in the gentle slope and passes area. In this paper, two small watersheds in the Linfen Basin of the Loess Plateau and the Sichuan Basin in the Yangtze River Basin were selected as study areas. SRTM DEMs of 30 m and 90 m are used in experiments. The results of hybrid flow direction algorithm are compared with the results of typical SFD and MFD algorithms. The divergent effect is apparently suppressed compared to multi-flow direction algorithms. The occurrence of unnatural parallel drainage lines is decreased compared to the results of single flow direction algorithms. And improvements of hybrid flow direction on DEMs of 30 m is better than 90 m datasets.

Current algorithms that deduce the drainage network from a digital elevation model (DEM) represented by a regular array of surface elevations share a fault: Unless the terrain is rugged, the derived water channels tend to flow in parallel lines along preferred directions engendered by the sampling grid orientation. We present a simple solution to the problem. A second difficulty is the presence of noise that creates artificial pits. We briefly describe a method which deals with pits in what we believe to be a more efficient manner for virtual memory environments than previous efforts. Our system has treated DEMs of nearly 9,000,000 pixels. We show how depth first search of the resulting drainage network permits segmentation of the DEM into basins by various criteria, analysis of stream‐sediment anomaly dilution profiles, improved hydrological models and other applications.

Current algorithms for computing contributing areas from a rectangular grid digital elevation model (DEM) use the flow‐routing model of O'Callaghan and Mark (1984), which has two major restrictions: (1) flow which originates over a two‐dimensional pixel is treated as a point source (nondimensional) and is projected downslope by a line (one dimensional) (Moore and Grayson, 1991), and (2) the flow direction in each pixel is restricted to eight possibilities. We show that large errors in the computed contributing areas result for any terrain topography: divergent, convergent, or planar. We present a new model, called digital elevation model networks (DEMON), which avoids the above problems by representing flow in two dimensions and directed by aspect. DEMON allows computation of both contributing and dispersal areas. DEMON offers the main advantage of contour‐based models (e.g., Moore et al., 1988), the representation of varying flow width over nonplanar topography, while having the convenience of using rectangular grid DEMs.