Supplemental data for Over and Sherwood (2025), Washover and washout locations and landcover classifications after hurricanes Florence, Dorian, Harvey, Ike, and Sandy in North Carolina, Texas, and New York.

GEOPACKAGE (Over_and_Sherwood_Supp_Extents.gpkg) NORTH CORE BANKS: Island platform region of interest (ROI) feature class (main.NCB_IslandPlatform_ROI) was created assess the landcover of the area adjacent to and affected by overwash and outwash events. The ROI aims to remove water (ocean) and back barrier marsh, which do not perform well in the unsupervised classification. The Hurricane Florence washover extents (main.NCB_Florence_Washover) were delineated by digitizing the extent of the interpreted sand class from an unsupervised classification (the number of classes is set to 3 with a minimum class size of 20 and sample interval of 10) of the 2018-10-07 orthomosaic (Ritchie and others, 2020) and clipping to the dune line in ArcPro (v 3.0.0). The Hurricane Dorian washout extents (main.NCB_Dorian_Washout) were delineated by the difference between NCB digital elevation models (Ritchie and others, 2022) from 2019-08-30 and 2019-09-12 following Sherwood and others (2023). Ponds in washout channels were hand digitized (main.NCB_Dorian_Washout_ponds) by J. Over, using the imagery from 2019-10-11, 2020-09-28, and 2022-10-20 (Kranenburg and others, 2022, 2023, 2024, 2025) that were turned into orthomosaics following Over and others (2021). SAN JOSE ISLAND: Individual washout channels were hand digitized by J. Over from 2017-08-29 georectified Google Earth imagery (main.SJI_Harvey_Washout). BOLIVAR PENINSULA: Hurricane Ike did not create individual washout channels, so two same sized ROI boxes were created next to each other on georectifed Google Earth Imagery to encompass the extent of washout that was replanted by the local government and an extent that was not remediated (main.BP_Ike_ROI). FIRE ISLAND: Washout extents were obtained from C. Kilheffer, and provided in main.FI_Sandy_Washover, to recreate the areas of vegetation recovery studied in Kilheffer and others (2019).

CLASSIFICATION DATA (Over_and_Sherwood_2025_Supp_VegetationCoverRaw.csv) Values are provided in the CSV by class, extent ID, and date. Percent cover is calculated by dividing the sum of the pixel count of vegetation classes by the pixel count sum of all classes and multiplied by 100. NORTH CORE BANKS: For each date (2018-10-08, 2019-08-30, 2019-10-11, 2020-09-28, 2022-10-20, 2024-11-06) an unsupervised classification was performed on the orthomosaic within the island platform ROI (main.NCB_IslandPlatform_ROI), the Hurricane Florence washover extents (main.NCB_Florence_Washover), and the Hurricane Dorian washout extents (main.NCB_Dorian_Washout) in ArcPro (v. 3.0.0) with 3 classes, a minimum class size of 20, and sample interval of 10. The three classes were visually interpreted into dense vegetation, sparse vegetation, and sand. Areas under the digitized ponds (when present), were assigned to water. A quality check was done with stratified random points using the Create Accuracy Assessment Points (n=200) and then the Compute Confusion Matrix Tool. For 2019-10-08 the overall accuracy (± 95% confidence interval (CI)) was 0.92±0.04, 2019-08-30 was 0.88±0.04, 2019-10-11 was 0.87±0.04, 2020-09-28 was 0.90±0.04, 2022-10-20 was 0.84±0.05, and 2024-11-06 was 0.88±0.04. SAN JOSE ISLAND: For each date (2017-03-25, 2017-08-29, 2020-10-31, 2022-03-23, and 2023-06-25) georectified Google Earth imagery was converted to a GeoTIFF. For each date an unsupervised classification was performed on the GeoTIFFs within the Hurricane Harvey washout extents (main.SJI_Harvey_Washout) in ArcPro (v. 3.0.0) with 3 classes, a minimum class size of 20, and sample interval of 10. The three classes were visually interpreted into vegetation, sand, and water. A quality check was done with stratified random points using the Create Accuracy Assessment Points (n=200) and then the Compute Confusion Matrix Tool. For 2017-03-25 the overall accuracy (± 95% CI) was 0.91±0.04, 2017-08-29 was 0.88±0.04, 2020-10-31 was 0.93±0.03, 2022-03-23 was 0.96±0.03, and 2023-06-25 was 0.94±0.04. Values are provided in the CSV by class, extent ID, and date. BOLIVAR PENINSULA: For each date (2008-05-19, 2010-01-08, 2013-04-19, 2018-01-02, and 2022-01-21) georectified Google Earth imagery was converted to a GeoTIFF. For each date an unsupervised classification was performed on the GeoTIFFs within the two Hurricane Ike ROIs (main.BP_Ike_ROI) in ArcPro (v. 3.0.0) with 3 classes, a minimum class size of 20, and sample interval of 10. The three classes were visually interpreted into vegetation, sand, and water. A quality check was done with random points using the Create Accuracy Assessment Points (n=100) and then the Compute Confusion Matrix Tool. For 2008-05-19 the overall accuracy (± 95% CI) was 0.92±0.05, 2010-01-08 was 0.91±0.05, 2013-04-19 was 0.96±0.04, 2018-01-02 was 0.97±0.03, and 2022-01-21 was 0.95±0.04. FIRE ISLAND: For each date (2017-10-01, 2019-09-19, and 2021-04-04) georectified Google Earth imagery was converted to a GeoTIFF. For each date an unsupervised classification was performed on the GeoTIFFs within the Hurricane Sandy washout extents (main.FI_Sandy_Washover) in ArcPro (v. 3.0.0) with 2 classes, a minimum class size of 20, and sample interval of 10. The two classes were visually interpreted into vegetation, and sand. A quality check was done with random points using the Create Accuracy Assessment Points (n=200) and then the Compute Confusion Matrix Tool. For 2017-10-01 the overall accuracy (± 95% CI) was 0.89±0.04, 2019-09-19 was 0.88±0.04, and 2021-04-04 was 0.89±0.04.