Enhancing hydrologic mapping using LIDAR and high resolution aerial photos on the Frances Marion National Forest in coastal South Carolina
Enhancing hydrologic mapping using LIDAR and high resolution aerial photos on the Frances Marion National Forest in coastal South Carolina
复制标题
使用激光雷达和高分辨率航空照片增强南卡罗来纳州沿海弗朗西斯马里恩国家森林的水文测绘
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
W. Hansen
中科院分区：
文献类型：
--
作者：
Andy Maceyka;W. Hansen
评估沿海海洋阶地特征内的水文一直是个问题，因为在低梯度地形和茂密的低地森林中很难确定流域边界和溪流细节。各种研究利用 USGS 地形等高线图（Hansen 2001，Eidson 等 2005）或沟壑山麓地形（James 等 2007）和马里兰沿海平原（Lang 等 2012）的光探测和测距（LIDAR）改善了水文细节。南卡罗来纳州 Huger 附近的土耳其溪子流域内的研究使用激光雷达和现场验证来估计 52.4 平方公里的子流域的大小，但 50 年的历史估计范围为 32.4 至 72.6 平方公里（Amatya 等人，2013 年）。土耳其溪是使用激光雷达评估的 21 个子流域之一，旨在用于覆盖 1,050 平方公里的弗朗西斯马里恩国家森林 (FMNF) 的规划修订。事实证明，激光雷达可以更准确地定义或定位河流网络和流域边界等许多事物，从而成为森林规划的宝贵资产。用于绘制土耳其溪地图的激光雷达数据是在 2009 年 2 月和 3 月获得的。土耳其溪的水流主要为 0.05-0.28 m3s-1（略低于 0.39 m3s-1 的 9 年平均值），因此大多数常年和间歇性溪流应该含有水，但小溪流和渗漏不太可能被注意到。高分辨率正射影像（ESRI 的世界影像、NAIP 2013 影像）也有助于图像解释。绘图过程采用了“平视数字化”和基于 DEM 的建模。这是一个数字化和重塑的迭代过程。从激光雷达衍生的 DEM 中去除了水文障碍，因此可以对流量进行建模。这是通过首先对通过代理清晰可见的流进行手动数字化来实现的，主要基于由于水吸收激光脉冲而丢失的激光雷达地面回波的线性性质。在平坦、潮湿的景观中，使用当前的地表建模方法常常会丢失这些有价值的信息。由于地形起伏较小和/或激光回波太少而无法正确定义地面与低植被或噪声回波，因此激光雷达衍生的 DEM 可能会“冲刷”区域内的河道。 “冲刷”效应是算法在没有其他激光返回（即水）的区域选择可用河岸或低植被激光返回的结果。在植被低且茂密的潮湿地区，地表误差尤其成问题。当这种“冲刷”发生时，单独使用当前基于 DEM 的建模可能很难对流网络进行建模。在这些具有挑战性的领域中使用当前基于 DEM 的方法时，需要进行大量工作来提供相对清晰的模型流路径。如果没有“清除”和定义的水文路径，流量累积模型常常会发生转向和河道接触松散。使用数字化线路工作“清除”流路径，以将流保持在其主通道中。通道不确定的区域，平面或垂直轮廓中的激光点云揭示了没有回波的区域，并且如果含有水，则可以识别溪流。尽管并不多产，但这些流动建模问题也发生在山前和山区，必须得到认识和处理。根据可识别的河道或连续的含水特征对溪流和沟渠进行数字化后，将流线刻录成 DEM 并进行重塑。流量建模工具只能识别一个通道，分析人员需要将其保留在主通道中。权衡图像和 LIDAR 得出的证据，考虑 DEM 统计数据设置为以当前显示范围（比例）刷新，使用离散颜色来分离高程细节，以及缺乏吸收 LIDAR 脉冲的水的返回，这些都需要在主通道网络中燃烧之前数字化精细的通道位置。识别具有辫状、蜿蜒至线性（沟渠）通道的特征地貌 1Andy Maceyka，Francis Marion & Sumter 国家森林 GIS 专家和南部地区 (R8) 遥感技术员，USDA 林务局，Francis Marion 和 Sumter 国家森林，哥伦比亚，南卡罗来纳州 29212 William Hansen，专业水文学家（退休的 USDA 林务局水文学家） Hansen Hydrology，所有者/运营商，列克星敦，南卡罗来纳州29072
Evaluating hydrology within coastal marine terrace features has always been problematic as watershed boundaries and stream detail are difficult to determine in low gradient terrain with dense bottomland forests. Various studies have improved hydrologic detail using USGS Topographic Contour Maps (Hansen 2001, Eidson and others 2005) or Light Detection and Ranging (LIDAR) in gullied piedmont terrain (James and others 2007), and the Maryland coastal plain (Lang and others 2012). Research within Turkey Creek subwatershed near Huger, SC used LIDAR and field verification to estimate the size of the 52.4 km2 subwatershed, but the 50-year history had estimates ranging from 32.4 to 72.6 km2 (Amatya and others 2013). Turkey Creek was one of 21 subwatersheds evaluated using LIDAR intended for the Plan Revision covering the 1,050 km2 Francis Marion National Forest (FMNF). LIDAR has proven to be a valuable asset to forest planning by more accurately defining or locating many things including stream networks and watershed boundaries. LIDAR data used to map Turkey Creek were attained in February and March of 2009. Streamflow in Turkey Creek was primarily 0.05-0.28 m3s-1 (somewhat below the 9-year average of 0.39 m3s-1) so most perennial and intermittent streams should contain water, but small streams and seeps are unlikely to be noticed. Highresolution ortho imagery (ESRI’s World Imagery, NAIP 2013 imagery) was also helpful for image interpretation. The mapping procedure employed both “heads-up digitizing” and DEM-based modeling. It was an iterative process of digitizing and remodeling. Hydrologic barriers were removed from the LIDAR-derived DEM so flow could be modeled. This was accomplished by first, hand digitizing streams that were clearly visible by proxy, based mostly on the linear nature of missing LIDAR ground returns due to the absorption of laser pulses by water. In flat, wet landscapes this valuable information is often lost using current methods to model ground surfaces. LIDAR-derived DEMs can “washout” stream channels in areas due to low topographic relief and/or too few laser returns to properly define ground versus low vegetation or noise returns. The “washout” effect is a result of the alogrithm selecting available stream bank or low vegetation laser returns in areas with no other laser returns (i.e. water). Errors in ground surface are especially problematic in wet areas with low, dense vegetation. When this “washout” occurs it can be difficult to model stream networks using current DEM-based modeling alone. When using current DEM-based methods in these challenging areas, a substantial amount of work is needed to provide a relatively clear path to model streams. Without a “cleared” and defined hydrologic path, the flow accumulation models often get diverted and loose channel contact. Stream paths are “cleared” using digitized line work to keep stream in its main channel. Areas of channel uncertainty, the laser point cloud in planimetric or vertical profile reveals areas with no returns and streams can be recognized if they contain water. Although not as prolific, these flow modeling issues also occur in the piedmont and mountains, and have to be recognized and dealt with. After digitizing the streams and ditches based on recognizable channels or continuous water bearing features, the stream lines were burned into a DEM and remodeled. The flow modeling tools identify only one channel, and the analyst needs to keep it in the main channel. Weighing the imagery and LIDAR derived evidence considering DEM statistics set to refresh with the current display extent (scale), using descrete colors to separate elevation detail and lack of returns from water absorbing the LIDAR pulses are all needed to digitize a refined channel location before burning in the primary channel network. Recognizing characteristic landforms with braided, meandering to linear (ditched) channels 1Andy Maceyka, Francis Marion & Sumter National Forest GIS Specialist and Southern Region (R8) Remote Sensing Technician, USDA Forest Service, Francis Marion and Sumter National Forest, Columbia, SC 29212 William Hansen, Professional Hydrologist (retired USDA Forest Service Hydrologist) Hansen Hydrology, owner/operator, Lexington, SC 29072