Assessing shorelines extracted from satellite imagery using coincident terrestrial lidar linescans

Numerous algorithms have been developed to facilitate the extraction of shoreline position from satellite imagery, including the CoastSat algorithm. Previous analyses comparing CoastSat satellite-derived shorelines (SDS) to morphological data have highlighted that there are site-specific errors in outputs that are likely related to concurrent wave runup conditions, but the data required to test this relationship are rarely available. Here, we present a novel comparison of lidar-derived runup and beach elevation data to CoastSat satellite-derived waterlines (SDW) extracted using two image sources (Sentinel-2 and PlanetScope) and two threshold algorithms (Otsu and weighted peaks). Results show that while SDW extracted using Otsu thresholds correlated better with lidar-derived waterlines (LDW), SDW extracted using the weighted peaks threshold were consistently positioned in the upper swash and therefore correlated better with a runup bulk statistic. Assigning the best-fit runup bulk statistic as the waterline elevation to weighted peaks SDW resulted in SDS with less scatter than the Otsu SDW due to the more consistent waterline elevation. Horizontal errors for the converted datum-referenced shoreline were lowest when SDW were converted to SDS using best-fit measured runup bulk statistics and a measured slope. However, for weighted peaks SDW from both image sources, assigning the best-fit parameterized runup bulk statistic and an average slope in the SDW to SDS conversion still reduced error by $\sim 20\text{\%}$ to $\sim 35\text{\%}$ when compared to the tidal elevation and average slope. These findings confirm that runup corrections can improve native SDS outputs, although the magnitude of the final shorelines error depended on the specific imagery product, local beach slope, threshold technique, runup parametrization, and chosen reference contour.