GRASS GIS, GMT and R scripts used to produce the figures in the peer-reviewed article by Polina Lemenkova. The scripts classify a Landsat 8-9 OLI/TIRS time series (February 2015, 2018, 2020, 2021, 2022, 2023) of the Saloum River Delta and coastal wetlands of Senegal, West Africa, comparing unsupervised k-means clustering with supervised Support Vector Machine (SVM) classification.
Published in: Earth 2024, 5(3), 420–462 DOI: https://doi.org/10.3390/earth5030024 Journal (open access): https://www.mdpi.com/2673-4834/5/3/24 SSRN: https://ssrn.com/abstract=4948887
- GRASS GIS shell scripts for raster import (r.import), clustering and classification (i.group, i.cluster k-means, i.maxlik maximum-likelihood), rejection-probability mapping, and Support Vector Machine classification (r.learn.train, r.learn.predict, SVC) from Python's Scikit-Learn library — one MaxLik and one SVM script per year.
- GMT script for the topographic study-area map.
- R (DiagrammeR) script for the methodological workflow diagram.
- Landsat metadata tables (LaTeX) and per-year clustering reports.
The LaTeX source (prose) of this article is in a separate repository: https://github.com/paulinelemenkova/svm-landcover-classification-senegal
Lemenkova, P. Support Vector Machine Algorithm for Mapping Land Cover Dynamics in Senegal, West Africa, Using Earth Observation Data. Earth 2024, 5(3), 420–462. https://doi.org/10.3390/earth5030024