Scholarlink

<p>Global digital elevation model (DEM) datasets, such as the Shuttle Radar Topography Mission (SRTM), are widely used for flood inundation modeling in developing nations. However, errors in these datasets can significantly impact the accuracy of analyses and applications. Attempts to improve DEM accuracy have used supplementary data like flood observations, vegetation data, and drainage networks, but these approaches are limited in data-scarce regions where such information is unavailable. To address this issue, this study proposes bias correction of DEM elevation values using ground truth data from field measurements and outlines a systematic framework to improve flood inundation mapping. Various ground truth sampling strategies are tested, and the minimum number of points required is identified. The framework is applied to the Brazos River, Texas, USA, using the HEC-RAS 2D model for a major flood event. Results demonstrate significant improvements in DEM accuracy following bias correction. The mean DEM error was reduced from 4.5&#xa0;m to 0.0&#xa0;m, with a notable decrease in error variance. The RMSE decreased by 55–70% across sampling strategies. Performance indices for inundation extent, F-index and C-index, improved by 47% and 60%, respectively, while the maximum depth simulation showed a 350% enhancement in Nash-Sutcliffe Efficiency (NSE), increasing from 0.13 to 0.59 when the corrected DEM was used. The results demonstrate that the proposed farmwork aids in significantly improving the vertical accuracy of global DEMs, enabling more precise hydrological simulations and flood risk assessments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Improving the usability of global SRTM DEM for reach-scale floodplain inundation mapping in data-scarce regions through bias correction

  • 期刊论文

  • 发表日期：

    2025年2月26日

查看全文
  • Ismail Jesna,
  • S. Murty Bhallamudi,
  • K. P. Sudheer

摘要

Global digital elevation model (DEM) datasets, such as the Shuttle Radar Topography Mission (SRTM), are widely used for flood inundation modeling in developing nations. However, errors in these datasets can significantly impact the accuracy of analyses and applications. Attempts to improve DEM accuracy have used supplementary data like flood observations, vegetation data, and drainage networks, but these approaches are limited in data-scarce regions where such information is unavailable. To address this issue, this study proposes bias correction of DEM elevation values using ground truth data from field measurements and outlines a systematic framework to improve flood inundation mapping. Various ground truth sampling strategies are tested, and the minimum number of points required is identified. The framework is applied to the Brazos River, Texas, USA, using the HEC-RAS 2D model for a major flood event. Results demonstrate significant improvements in DEM accuracy following bias correction. The mean DEM error was reduced from 4.5 m to 0.0 m, with a notable decrease in error variance. The RMSE decreased by 55–70% across sampling strategies. Performance indices for inundation extent, F-index and C-index, improved by 47% and 60%, respectively, while the maximum depth simulation showed a 350% enhancement in Nash-Sutcliffe Efficiency (NSE), increasing from 0.13 to 0.59 when the corrected DEM was used. The results demonstrate that the proposed farmwork aids in significantly improving the vertical accuracy of global DEMs, enabling more precise hydrological simulations and flood risk assessments.