Evaluation of Sentinel-2 time-series for mapping floodplain grassland plant communities

Sentinel-2 time-series data significantly enhances the accuracy of mapping floodplain grassland plant communities compared to single-date or single-band imagery. A study at Mont-Saint-Michel Bay, France, demonstrated that using a full annual Sentinel-2 time-series with a support vector machine (SVM) classifier achieved an overall accuracy of 0.78, outperforming random forest (RF) classifiers and single-date/single-band datasets. The findings highlight the importance of temporal resolution in discriminating plant communities, as accuracy dropped by 34% when using single-date images and 12% with single-band data. This research underscores the potential of Sentinel-2 for biodiversity monitoring in temperate floodplains.

Rapinel, S., Mony, C., Lecoq, L., Clément, B., Thomas, A., & Hubert-Moy, L. (2019). Evaluation of Sentinel-2 time-series for mapping floodplain grassland plant communities. Remote Sensing of Environment, 223, 115–129. https://doi.org/10.1016/j.rse.2019.01.018