Marsh habitat change analysis for the Point Aux Chenes and Grand Bay Estuaries in Mississippi and Alabama from 1848 to 2022

Topographic sheets (t-sheets) for 1848 (T00243 and T00273) and 1957/58 (T10744, T10745, and T10752) were added to ArcMap and projected into the North American Datum of 1983 Universal Transverse Mercator Zone 16N (NAD83 UTM 16N). Data was checked to ensure positional accuracy with other previously published data sources. During this check, it was discovered T00273 was slightly shifted from T00243 due to an error within the published NOAA data. Using the ArcGIS Pro georeferencing toolbar, the T00273 t-sheet was georectified using the T00243 t-sheet as a reference. Each t-sheet contained lines representing the shoreline and upland boundary that were heads up digitized within ArcGIS Pro using the Editor toolbar at a scale of 1:1500.

Worldview 2 imagery from 20191116 (YYYYMMDD) and 20220511 were 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 common projection system of Universal Transverse Mercator World Geodetic System of 1984 datum (UTM zone 16N, WGS 1984). 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 meter were removed. A methodology similar to that described by Maglione and others (2014) was used to generate vector lines from WorldView (WV) images. 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 0.65) were classified as upland, low values (below 0.21) were classified as water, and in between values (0.21 to 0.65) were classifed as marsh. Using ArcGIS Pro expand and shrink tools, isolated pixels were removed from the wetland-upland raster. In ArcGIS Pro, 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 Polynomial Approximation with Exponential Kernel (PAEK) algorithm and a 2-meter smoothing filter was used. During a quality review, it was discovered that there were several single cells that were classified as upland but were not visibly upland on the WorldView imagery. Collections of these cells with a circumference less than 100 meters were removed as they often were misclassified cells and not upland. Lines were then projected into NAD83 UTM 16N.

Within ArcGIS Pro, the shorelines and upland lines for the 1848, 1957, and 2022 were added to a map. A single bounding box extent was created that encompassed all three dates and intersected the furthest left and right ends of shorelines and upland lines. This bounding box was created so all three dates have the same final raster extent. The bounding box was then copied into each dated shoreline/upland file. The ends of the shorelines and upland lines were checked to ensure they crossed the left and right edges of the bounding box. The line files were converted into polygons using the ArcGIS Pro tool “polyline to polygon”. The polygons were checked to ensure the classification field containing water, marsh, and upland information was still present. A new classification type was added for polygons that represented water as “0”, marsh as "1", and upland as "2". The polygon was then converted into a raster using the “Polygon to raster” tool with a cell assignment type of maximum combined area and a cell size of 15 meters. The 1848 raster cell values were reclassed to water=1, marsh=3, and upland=7. The 2022 raster cell values were reclassed to water=10, marsh=30, and upland=70. Using the raster calculator, the 1848 raster cell values were added to the 2022 raster cell values. A new field was added to the resulting raster with text codes for each new cell value. For example, if a cell was water in 1848 (cell=1) and marsh in 2022 (cell=30) then the new cell would have a value of “31” (1+30), and the change type of "water to marsh" was added to a field was called "habitat_change". The 1957 image was given 2 classification codes since it was both the last date (1848-1957) and the first date (1957-2022) in analyses. As a result, for the first date (1957-2022) run the classification was water=1, marsh=3, and upland=7 and for the last date (1848-1957) run the values were changed to water=10, marsh=30, and upland=70. The raster calculator was then used to calculate the sum of 1848+1957 and 1957+2022.