Data from: Mapping coastal forest retreat using convolutional neural networks and satellite imagery
Data files
Sep 03, 2026 version files 45.77 MB
-
pred_Landcover1985.tif
22.89 MB
-
pred_Landcover2021.tif
22.89 MB
-
README.md
1.19 KB
Abstract
This dataset represents the classified land cover data for 1985 and 2021, generated from the study titled "Mapping Coastal Forest Retreat Using Convolutional Neural Networks and Different Satellite Imagery", published in the PLOS ONE journal (DOI: 10.1371/journal.pone.0357346).
Seven land cover classes are represented: -1 (Invalid pixels), 0 (Marsh), 1 (Shrub), 2 (Ghost Forest), 3 (Forest), 4 (Cultivated), and 5 (Water). The dataset is provided to support analyses of land cover composition, coastal forest dynamics, and changes associated with environmental and climate-related processes.
Dataset DOI: 10.5061/dryad.w9ghx3g5w
Description of the data and file structure
The land cover raster data include seven land cover classes: -1 (Invalid pixels), 0 (Marsh), 1 (Shrub), 2 (Ghost Forest), 3 (Forest), 4 (Cultivated), and 5 (Water). The dataset is provided to support analyses of land cover composition, coastal forest dynamics, and changes associated with environmental and climate-related processes.
The land cover classes were generated using summer spectral, winter spectral, and annual phenological indices, as well as topographic and multispectral indices, as described in the Methods section of the journal article accompanying this published dataset. The original datasets used to generate these land cover data are publicly available through the USGS EarthExplorer platform (https://earthexplorer.usgs.gov/).
Files and variables
File: pred_Landcover2021.tif
Description: 2021 Classified map
File: pred_Landcover1985.tif
Description: 1985 Classified map