USJRB PLNT CMMNT2022

This is the 2022 ensemble classification map for Upper St Johns River Basin Plant Communities. The map was created based upon the classification of a 0.3-m and 8-band pansharpened WorldView-2 image collected on 2 March 2022 using three contemporary machine learning classifiers: Artificial Neural Network (ANN), Support Vector Machine (SVM), and Random Forest (RF). The ensemble analysis combined the outputs of ANN, SVM and RF to generate a more robust classification. The generation of this map used eCognition Developer 10.0 for image segmentation, R script for machine learning and classification, and ArcGIS Pro for training sample selection and map product generation. For the BCWMA area, a smaller scale (150) was used in eCognition to create image object due to a higher heterogeneity of this area, while other five regions were created using a scale of 300. The map has an overall accuracy above 80% for classifying 22 communities/land cover types in terms of a combination of reference samples collected in the field and interpreted from WV-2 using 10-fold cross validation method. Anthropogenic and Levee/Road classes were manually mapped.