地理信息系统文献翻译——摘要

在本篇论文中，我们旨在解决空间分辨率的变化产生的影响，流系统的几何形状的地形类型以及使用自动化的地形特征提取方法从数字高程模型提取流域。对三个地形类型的五个不同的分辨率进行了分析，由此获得使用几个标准水文指标比较的流的载体。首先，我们使用ArcGIS的水文分析工具证明排水和流域的产生，并在跨地形类型测试时，发现特定地形参数的重大的结构差异。出乎意料的是，虽然平坦的地势有一些非常明显的规模行为，但是只有适当地形高低的混合地形可以比平坦或丘陵地形更好地显示变化。其次，随着单元尺寸的分辨率变小，在许多地形参数中有一个可衡量的增加值，这足以明确指出在参数中有明显的规模依赖。最后，我们证明了激光雷达数据表明这种变异效果的增加。该分析结果可能会产生新的算法和方法，利用激光雷达数据实现从数字高程模型进行更高精度的特征提取。

引言

上部的景观地形表面可以划分成一组非重叠的，详尽的多边形区域。这些区域与常见的描述性术语的景观特点一样，其中的界限往往被称为面网。这些功能包括湖泊，排水管线，山脊，山谷，山坡，鞍部和凹陷处。未来的地理信息系统中，这些以及其他的功能，而不是地图瓦片，将在所谓的地理信息系统的对象模型中成为属性存储和数据检索的基础。这些功能可以通过处理数字高程模型被自动地提取，同时使用现成的程序和自定义算法。自动标签识别这些功能不仅是进一步分析的重要步骤，也是为了建立一个一致的地图功能本体和后续任务，例如地图搜索功能注释。天然地形特征已经通过计算机从数字高程模型对下坡流的积累这一方法完成提取，并阈值这些值来提取排水和脊载体（人名不用翻译，2000）。数字高程模型精度和分辨率的一个显著的改善已经导致激光雷达测绘地形的广泛使用（NRC，2007）。精度和分辨率的提高已经推动了对描述这些地形特征更加精确的需要，让他们也可以唉地图本体中与其他功能，如道路和文化功能合二为一。在这项研究中，我们审查了使用更高数字高程模型分辨率对未来提取过程产生的影响。我们使用一个提取过程，但在不同的空间分辨率和三种类型的数字高程模型之间测试其效果：平坦，起伏地形......

ABSTRACT 

In this paper, we address the impact of changing patial resolution and terrain type on the geometry of stream systems and drainage basins as extracted from digital elevation models (DEMs) using automated terrain feature extraction methods. Five ifferent resolutions for each of three terrain types were analyzed to derive stream vectors for comparison using several standard hydrological metrics. First, we demonstrate that the generation of drainage and watershed features using ArcGIS’s ArcHydro tools reveals major structural geometric differences in specific terrain parameters when tested across terrain types. Unexpectedly, mixed terrain with only moderate relief shows more variance than flat or hilly terrain, though flat terrain has some very distinct scale behaviors. Second, as resolution in terms of cell size gets smaller, there is a measurable increase in many terrain parameters, sufficient to definitively state that there are significant scale dependencies in the parameters. Finally, we indicate that LiDAR data demonstrates this increase in variance effect. The results of this analysis may lead to new algorithms and approaches that use LiDAR to achieve higher levels of accuracy in terrain feature extraction from DEMs. 

INTRODUCTION 

The upper terrain surface of the landscape can be divided into a set of non-overlapping, exhaustive polygonal regions that characterize landscape features with common descriptive terms, the boundaries of which are often called the surface network. These features include lakes, drainage lines, ridges, valleys, hillslopes, saddles, and depressions. 

In future geographic information systems these and other features, rather than map tiles, will be the basis for storing attributes and searching data, in the so-called object model for GIS. The features can be extracted automatically by processing digital elevation models (DEMs) using both off-the-shelf routines and custom algorithms. Identification of these features for automated labeling is an important step not only for further analysis, but also for building a consistent ontology of map features and for subsequent tasks such as map search and feature annotation. Natural terrain features have been extracted by computing downslope flow accumulation from a DEM, and thresholding these values to extract drainage and ridge vectors (Wilson and Gallant, 2000)......