El Salvador
P. Paz, J. Perez, S. Leveron, J. Tello

The Terra-i team together with CRS El Salvador under the Raices project carried out a virtual workshop through the teams platform to technicians from the Ministry of Environment and Natural Resources, CARITAS, Universidad El Salvador, CENTA, about the Mapping of land cover using remote sensors and open source tools such as GEE, SEPAL and QGIS- Plugin Semi Automatic Classification.

During 3 days, the procedure to classify land cover was carried out using satellite images and automated methods in the Department of Chalatenango in El Salvador. The methodology consisted of four stages described in Figure 1, first, the input data collection was done using high-resolution images, then the pre-processing of satellite images was done, followed by the evaluation of different unsupervised and supervised methods, finally the evaluation of these through the confusion matrix and evaluation metrics.

Figure 1. Open source tools used in each of the steps to do the land cover classification


The K-MEANS, RF, SVM, CART classifiers were evaluated through the GEE platform using sentinel-2 images. The spectral classifiers Maximum Likelihood, Minimum Distance and Spectral Angle were evaluated through the QGIS Semi Automatic Classification plugin. These were used to identify six macro classes of land cover corresponding to: bodies of water, forest, crops, bare soils, pastures, and urban areas.

Figure 2. Land cover classification using Sentinel 2A and Supervised methods in GEE

Figure 3. Land cover classification using Maximum Likelihood method on QGIS Plugin Semi Automatic Classification.


Participants were able to experiment with the different land cover classifiers with each of the free tools, some based on cloud processing such as GEE and SEPAL, and a desktop tool such as the QGIS Semi Automatic Classification plugin, evaluating the optimal results for each coverage, determining the advantage of each method and tool.

The workshop was the first approach to use open source tools for satellite image processing and land cover analysis in the region.

"The land cover mapping would be a basic input to be able to make site diagnoses, project development, multi-temporal evaluations among many other things, including land use planning."

Workshop participant Dario Chavez, from Universidad El Salvador


The Terra-i team of the Bioversity International Alliance and CIAT thanks the participants and all those involved in the execution of the Workshop. This workshop is part of the Agricultural Landscape Restoration Initiative - RAÍCES Ahuachapán, funded by CRS El Salvador.

Important articles Featured category

Important articles News category