Drainage network analysis is fundamental to understanding the characteristics of surface hydrology. Based on elevation data, drainage network analysis is often used to extract key hydrological features like drainage networks and streamlines. Limited by raster-based data models, conventional drainage network algorithms typically allow water to flow in 4 or 8 directions (surrounding grids) from a raster grid. To resolve this limitation, this paper describes a new vector-based method for drainage network analysis that allows water to flow in any direction around each location. The method is enabled by rapid advances in Light Detection and Ranging (LiDAR) remote sensing and high-performance computing. The drainage network analysis is conducted using a high-density point cloud instead of Digital Elevation Models (DEMs) at coarse resolutions. Our computational experiments show that the vector-based method can better capture water flows without limiting the number of directions due to imprecise DEMs. Our case study applies the method to Rowan County watershed, North Carolina in the US. After comparing the drainage networks and streamlines detected with corresponding reference data from US Geological Survey generated from the Geonet software, we find that the new method performs well in capturing the characteristics of water flows on landscape surfaces in order to form an accurate drainage network.

排水网络分析是了解地表水文特征的基础。基于高程数据，排水管网分析通常用于提取关键水文特征，如排水管网和流线。受基于栅格的数据模型的限制，传统的排水网络算法通常允许水从栅格网格向 4 或 8 个方向（周围网格）流动。为了解决这一限制，本文介绍了一种新的基于矢量的排水网络分析方法，该方法允许水在每个位置周围以任何方向流动。该方法是通过光检测和测距 (LiDAR) 遥感和高性能计算的快速进步实现的。排水网络分析是使用高密度点云而不是粗分辨率的数字高程模型 (DEM) 进行的。我们的计算实验表明，基于矢量的方法可以更好地捕获水流，而不会由于 DEM 不精确而限制方向的数量。我们的案例研究将该方法应用于美国北卡罗来纳州罗文县的流域。将检测到的排水管网和流线与由Geonet软件生成的美国地质调查局的相应参考数据进行比较后，我们发现新方法在捕捉景观表面水流特征以形成准确的排水管网方面表现良好。