Fig. 7: Assessment of the impact on the accuracy of using different satellite sensors and adding of a wave setup correction.
From: Benchmarking satellite-derived shoreline mapping algorithms
a The SDS time-series for the 3 algorithms that use individual satellite images were used to compare the accuracy of the shorelines derived on Landsat and Sentinel-2 images. The full and hatched boxplots show the horizontal errors associated with the Landsat and Sentinel-2 time-series, respectively, for the 3 instantaneous shoreline algorithms (CoastSat, SHOREX, CASSIE). At Torrey Pines, as the ground truth data only spans to the end of 2016, only 1 year of Sentinel-2 data could be evaluated. The value of the median is indicated, and the whiskers are set at 1.5xIQR. Note that the y-axis was stretched for Truc Vert to accommodate for the larger errors. b A wave setup correction, based on Stockdon et al.41, was applied to the Landsat tidally corrected SDS time-series. The full and hatched boxplots show the horizontal errors associated with the tide-only and tide + wave setup time-series, respectively, for the 3 instantaneous shoreline algorithms (CoastSat, SHOREX, CASSIE). The error metrics describing the distributions in a and b are presented in Tables 4 and 5, respectively.
