Identifying Potentially Resolvable National Hydrography Dataset Waterbodies and Flowlines from Landsat Imagery

Remotely sensed data from satellite platforms (e.g. Landsat and Sentinel) are increasingly being used to monitor water quality and quantity through time. However, remotely sensed data are not of sufficient resolution to monitor all waterbodies. Waterbodies where the maximum width of surface water extent is less than a sensors spatial resolution and waterbodies where vegetation (either terrestrial or aquatic) impedes the overhead view of water cannot be reliably observed remotely. Knowing which waterbodies contain satellite-observable water (i.e. resolvable waterbodies) is the first step in determining if remote sensing is an appropriate monitoring method. Using the Global Surface Water Extent Dataset (GSWE) from the Global Surface Water Explorer to specify where water was observed by Landsat from 1984 to 2020, we (the USGS Idaho Water Science Center as part of the Next Generation Water Observing System Project) developed a workflow to identify waterbodies and flowlines from the High-Resolution National Hydrography Dataset with added attributes (NHDPlus HR) that are potentially resolvable from Landsat imagery for the United States (excluding Alaska). The workflow determines where the extent of NHDPlus HR water features (waterbodies and areas) overlaps the maximum extent of Landsat-observed water (as defined by GSWE), then removes edge pixels to determine if any pure water pixels are observable by Landsat within a water features extent. Each pure-water pixel within NHDPlus HR area features (i.e. rivers represented by polygons) was associated with the nearest NHDPlus HR flowline. The result identifies flowlines and waterbodies (as defined by NHDPlus HR) that contain at least one estimated pure-water pixel from the GSWE and are thus potentially resolvable. The resulting dataset can be linked to the NHDWaterbodies and NHDFlowline layers to identify waterbodies where remote sensing may be a suitable monitoring method.