Coastal Features Extracted from Landsat Satellite Imagery, Sabine Pass to Bay Coquette, Louisiana, 2013-2024

The regional AOI, which was derived from the State of Louisiana’s Barrier Island Comprehensive Monitoring (BICM) Program habitat mapping extents (Enwright and others, 2020), was subset into three scene-specific AOIs for processing and analysis: WRS-2 path 22 row 40 (dr22_2240), WRS-2 path 23 row 40 (dr22_2340), and WRS-2 path 24 row 39 (dr22_2439).

For each image acquisition date, Collection-2, Level-2 surface reflectance (SR; reflective bands), surface temperature (ST; thermal infrared [TIR] bands), and SR-derived NDVI images were downloaded from the EROS ESPA On Demand Interface (https://espa.cr.usgs.gov/).

All images were batch-processed using Spatial Model Editor in ERDAS IMAGINE. The SR bands were stacked to create 8-band multispectral images and clipped to the scene-specific AOI. From these composite images and the corresponding ST images, two additional spectral indices, mNDWI (Xu, 2006) and NBLI (Li and others, 2017), were calculated.

Land-cover classification was modified from the workflow described by Bernier and others (2021) using single (mNDWI), multilevel (2 thresholds per image; NBLI), or iterative (NDVI) thresholding of spectral indices and applying a rule-based classification scheme. First, Otsu's method (Otsu, 1979) for automatic histogram thresholding was applied to mNDWI images to create a binary "land"-water raster for each image acquisition date. Second, water and urban areas (derived from the BICM “structure” class [Enwright and others, 2020]) were masked from NBLI and NDVI images, and Otsu's method was applied to the masked images to create binary bare earth-"unclassed" and vegetated-"unclassed" rasters, respectively. For the WRS-2 path 23 row 40 (dr22_2340) scene-specific AOI only, the downloaded ST images included 2 large areas of missing data east of Freshwater Bayou Canal, which are the result of missing Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Emissivity Dataset (GED) input data (see https://www.usgs.gov/landsat-missions/landsat-collection-2-surface-temperature-data-gaps-due-missing-aster-ged for more information) necessary for ST product generation. Without ST data, NBLI could not be calculated for these pixels and any land-cover classification relying on NBLI input is invalid. Therefore, a mask was created from the ST “NoData” pixels and included in the land-cover classification. The following rule-based classification was then applied with each successive step applied to any unclassed pixels from the previous steps, where T represents the single Otsu threshold, T1 and T2 represent the first and second multi-Otsu thresholds, respectively, and Ti represents the second iteration of Otsu thresholding for each spectral index: If mNDWI > T then class (1) = water If ST = NoData then class (2) = land (undifferentiated) If NBLI > T2 then class (3) = bare earth (sand) If NDVI > Ti or (NDVI < Ti and NBLI < T1) then class (4) = vegetated If T1 < NBLI < T2 and NDVI < Ti then class (11) = intertidal Finally, the binary land-cover images were converted to thematic rasters, merged, single-pixel "clumps" were removed using a 3x3 majority filter, and a standard colormap was applied to create a final land-cover raster dataset for each image acquisition date. The resulting land-cover rasters use the naming convention YYYYMMDD_lc##_AOI_lcr_ce.img, where YYYYMMDD denotes the image-acquisition date (4-digit year, 2-digit month, 2-digit day), lc## denotes Landsat 8 (lc08) or Landsat 9 (lc09) image source, AOI denotes the scene-specific AOI, and lcr_ce are process step abbreviations where lcr indicates the "raw" land cover files that were created by thresholding the spectral indices were merged following the rule-based classification and ce indicates that single-pixel "clumps" were removed. All steps were batch-processed using the Image Processing toolbox in MATLAB (Otsu thresholding and binary image creation) or Spatial Model Editor in ERDAS IMAGINE (spectral index masking and generation of classed land-cover rasters).

Vector shoreline (representing the boundary between open-water areas and adjacent non-water land cover pixels, including intertidal areas) and the sand extents were extracted by contouring the mNDWI and masked NBLI images using the calculated Otsu thresholds. Shoreline vectors were manually cleaned to remove interior water bodies and contours representing extents of less than 4 connected pixels in the landcover rasters. The resulting shorelines include only the seaward shoreline for mainland land areas or the sea and back-barrier shorelines for barrier islands; however, interior shorelines (for example, along fluvial or tidal inlets or complex wetland shorelines) that are connected to the sea shoreline were not manually clipped and removed. Sand vectors were manually cleaned to remove interior mainland (non-beach) areas and contours representing sand extents of less than 4 connected pixels in the landcover rasters. The resulting shapefiles (one per image acquisition date) were merged into 2 shapefiles (one each for shoreline and sand features) for each scene-specific AOI. For the 37 datasets that were clipped to either the eastern or western part of the scene-specific AOI to exclude cloud cover, the vectors sand and shoreline files were also clipped to the same extent. The resulting sand (sandext) and shoreline (shrln) vector shapefiles use the naming convention AOI_shrln.shp or AOI_sandext.shp, where AOI denotes the scene-specific AOI. Contouring was batch-processed using Python 3 Jupyter Notebooks in ArcGIS Pro.