Shorelines, shorepoints, and transects with rates for the Point Aux Chenes and Grand Bay Estuaries in Mississippi and Alabama from 1848 to 2023

Shorelines for 1848-2017 were downloaded from the Terrano and others (2018) data release as shapefiles (.shp) that were projected into the North American Datum of 1983 Universal Transverse Mercator zone 16 North (NAD83 UTM 16N). These shorelines were already in the necessary format for the shoreline change analysis and were not modified for use in this study. Since these shorelines originated from several internet sources or were created by the USGS, please refer to Terrano and others (2018) for data sources, processing steps, and attribute accuracy reports related to these shorelines.

From 2013 to 2023, coastal managers surveyed shoreline positions using Real-time Kinematic (RTK) GPS at eleven locations in GNDNERR. Shorelines from 2013-2017 were previously published in Terrano and others (2018). Shorelines from 2018-2023 were collected as part of this study using similar methods to the 2013-2017 data. GPS data were collected using a Trimble R8 Model 3 Global Navigation Satellite System (GNSS) and TSC3 data collector from 2013 to 2018, or a Trimble R10 GNSS system and TSC3 data collector from 2018 to 2020. Each was attached onto a 2 m graphite rod with a mounted foot to obtain both horizontal and vertical shoreline position. The positional accuracy of Trimble R8 Model 3 GPS points was ± 10 millimeters (mm) + 1 parts per million (ppm) root mean square (RMS) horizontal error and ± 20 mm + 1 ppm RMS vertical error. The horizontal error of the Trimble R10 GPS points was ± 8 mm + 0.5 ppm RMS and vertical error was ± 15 mm + 0.5 ppm RMS. The points were collected roughly 5 to 10 meters apart along the vegetation-water boundary, which typically represented the top of an erosional scarp. In instances where an erosional scarp was not visible, the most suitable shoreline position based on dense shoreline vegetation was mapped. Shorelines were mapped at GNDNERR quarterly erosion sites or at USGS study sites and do not have full study area coverage. GPS points were then converted into a shapefile. The points were converted into vector shorelines using the XTools Pro (v.17) "polylines from points" tool. The location between each point was interpolated, so the specific shoreline location between each point may not represent the actual shoreline position.

WorldView 2 imagery from 20220511 (YYYYMMDD, image ID number WV220220511164242) was downloaded from Maxar’s G-EGD. The images were radiometrically and atmospherically corrected and then pansharpened using ERDAS IMAGINE 2020 (version 16.6.0) to obtain measures of ground reflectance. The output image was reprojected into the World Geodetic System of 1984 Universal Transverse Mercator zone 16 North coordinate system (WGS84 UTM zone 16N). The image was then co-registered to high-resolution aerial imagery (NAIP) using the AutoSync-workstation in ERDAS Image pixels. First, an NAIP image mosaic was created for an extent larger than the WV image coverage. Ground control points coincident on both images were used to adjust the WV image to the corresponding location on the NAIP. Control points with an error value greater than 1 were removed. A methodology similar to that described by Maglione and others (2014) was used to generate vector shorelines from WorldView (WV) images. All WV images were classified into land-water rasters using tools within ArcGIS Pro. Then, normalized difference vegetation index (NDVI) was calculated using WV band 5 in the visible red spectrum (RED) and band 7 in the near-infrared spectrum (NIR1) using the following formula: NDVI=(NIR1−RED)/(NIR1+RED). High values (above 0.2) are classified as vegetation, low values (less than -0.2) as water, and in between values (-0.2 to 0.2) as soil. A value of 0.21 provides an adequate representation of the shoreline (wetland-water boundary) and that value was used to reclassify the NDVI to a land-water raster. The land-water raster was processed to reduce isolated pixels by using expand-shrink procedure. The resulting image classified vegetated shorelines. Due to the high reflectance of beach and urban areas a second classification using band 8 (NIR2) with a threshold value of 5000 was used. Values above 5000 were considered urban or beach. The beach/urban classification was merged with the vegetated land classification to create a binomial land-water raster. In ArcMap, the classification boundaries were smoothed to remove any extraneous cells missed in the expand and shrink steps. The raster was then converted into polygons and the polygons were converted into polylines. To smooth the ridged cell edges, the lines were smoothed using the Polynomial Approximation with Exponential Kernel (PAEK) algorithm at a 2-meter smoothing filter. A final quality control was conducted and any shorelines that were not in the study area or shorelines improperly classified (such as nearshore sand shoals) were removed or manually edited. A second USGS GIS user then conducted a second quality control check and updates were made as necessary. Shorelines were projected into NAD83 UTM 16N. The shoreline was run through the Arc Pro "Generalize" tool set to a filter of 1 meter to reduce the number of shoreline vertices.

The vectorized shorelines that were produced in the previous process steps were compiled in ArcMap. These vectorized shorelines were manually edited in ArcMap so that only the desired shorelines were included in the final dataset. Once shorelines were edited and checked, all dated shorelines were merged into a single shapefile for inclusion in this data release, Shorelines_1848_2023.shp. Attributes were edited to include information regarding the date and source of the original data. The compiled shorelines were processed using the Analyzing Moving Boundaries Using R (AMBUR) statistical package for R (version 4.0.4) to add required missing attributes and calculate shoreline change rates. AMBUR performs shoreline change analysis on vector digital shorelines (Jackson, 2010). All necessary attribute fields must be present in order to run various AMBUR functions. The following fields are required for boundaries: "Id", "Date_", "Accuracy", "SHORE_LOC", "CLASS_1", "CLASS_2", "CLASS_3", and "GROUP". In order to determine if all the required fields were present in the merged boundary datasets, an AMBUR sub-routine (ambur.check) was used to check all shoreline shapefiles for missing attribute fields. During the shoreline change analysis, AMBUR also creates shorepoints. A shorepoint was created at each transect and shoreline intersect point. Several transects intersected shorelines that doubled back or intersected both sides of an island. In these instances, multiple shorepoints were created for the same date on the same transect. To remove the duplicate points, the shorepoint file was then run through an R code to remove all shorepoints after the first intersection for each date. For more information about the AMBUR shoreline requirements see the AMBUR user manual (Jackson, 2010). Some fields created by AMBUR ("SHORE_LOC", "CLASS_1", "CLASS_2", "CLASS_3", and "GROUP") are required and have null values as placeholders. Several islands within the study area were not used in the shoreline change analysis as they were too far offshore but were retained in the shoreline data set. Null fields were left in the attribute table as they are required by AMBUR and are needed to replicate this study.

Transects were generated by buffering the shoreline dataset produced in the previous step to create inner (land-based) and outer (water-based) baselines. The baselines form the start and end point for shoreline-perpendicular transects. Baselines were input into the AMBUR statistical package for R (using R version 4.0.4). AMBUR has several tools, which utilize shoreline parallel baselines to generate transects that are generally perpendicular to the shoreline. Transects were constructed at a sampling distance of 50 meters. Length varied from 1000 to 3000 meters, depending on location, and was selected to cover all dated shorelines. Once transects were created, the "Filter Transects" tool was used to adjust and even out the spread of transects. Where shorelines experience sharp bends, such as in small bays and narrow spits, transects may fall at a non-perpendicular angle. Small islands located within 50 meters of the main island boundary were considered a part of the barrier island system and were included in the analyses. Final transects were checked and edited in ArcGIS Pro, if necessary, in order to improve analysis results and improve shoreline coverage. Transects were manually edited to reduce errors in the analyses. Shoreline points and final statistical analyses were completed in AMBUR to generate the shoreline change rates. The following analysis parameters were used: first intersection (if transect intersects the same shoreline more than once, then by default it selected the first intersection), confidence level 95 (confidence level for the linear regression statistics), unit label m (the units of measure for the map is for Universal Transverse Mercator and in meters), analysis type is advanced (advanced includes additional statistics for a robust linear regression), and time unit for rates is yr (utilizes years for calculating rates of shoreline change). For more information on the AMBUR program, refer to Jackson (2010).