Overview This repository contains a script for processing satellite imagery to classify land use and land cover in Malta. The script loads satellite images, clips them to the region of interest, prepares training data, performs supervised classification using Random Forest, and visualizes the results.
The script requires the following R packages:
raster: For raster data manipulation rgdal: For bindings to GDAL for spatial data operations sp: For handling spatial data ggplot2: For generating plots sf: For handling simple features and spatial vector data caret: For machine learning algorithms ranger: For efficient Random Forest computation rasterVis: For enhanced raster visualization Please ensure these packages are installed before running the script.
The script utilises two sets of satellite imagery reflecting dry and wet conditions. It also uses a shapefile for the Malta region to clip the satellite images and align the training data. Ensure the data is correctly placed in the directory as specified in the script.
To run the script, simply load it into your R environment and execute. The script is well-commented and divided into sections for ease of understanding and modification.
The script outputs:
Clipped raster images of Malta for dry and wet conditions. A data frame with prepared training data. Trained Random Forest models for both dry and wet conditions. Confusion matrices and accuracy assessments for the models. Predicted land cover classification maps for both conditions. Enhanced visualizations of the classified rasters with color legends. Contributing Contributions to the script are welcome. Please fork the repository and submit a pull request with your suggested changes.
This project is licensed under the [LICENSE NAME]. See the LICENSE file for more details.
For any questions or support, please contact Ecostack Innovations at https://www.ecostackinnovations.com/.