Phenology-based classification of Sentinel-2 data to detect coastal mangroves

Precise categorization of mangrove forests with medium spatial resolution satellite data is challenging and occasionally yields mixed outcomes. The available methods to estimate mangrove vegetation cover using moderately high-resolution images lack differentiation between mangrove and homestead vegetation. Mangrove vegetation displays a range of responses across the phenological cycle at different wavelengths of an optical sensor. By taking advantage of this principle, the study applied some mangrove and non-mangrove VIs as predictor variables sourced from monthly Sentinel-2 data. These variables were grouped by individual VIs and fed into the random forest algorithm to derive phenology-based classification. A suitable month for thresholding mangroves across different VIs was also ascertained. Results indicated that phenology-based classification with three classes was more accurate (95% overall accuracy) than threshold-based or WorldCover v100 classifications. MI and MVI layers from December image performed better in discerning mangroves. Findings have important implications in separating mangroves from other coastal vegetations.

Automated summary

Accurate classification of mangrove forests using medium spatial resolution satellite data is difficult, and commonly results in mixed outcomes. This study utilized the responses of mangrove vegetation at different wavelengths, and employed random forest algorithm with extracted vegetation indices to achieve phenology-based classification of mangroves. The optimal thresholding month for mangroves was also determined. Results showed that the phenology-based classification with three classes was more accurate than threshold-based or WorldCover v100 classifications, with December image performing better in discerning mangroves. These findings have important implications for separating mangroves from other coastal vegetations.