Fig. 5: Performance of the hybrid transect classifier and importance of the used features.

Left-hand panel a: overall accuracy for hybrid transect classifier when validating against the coastal transect types of sandy, muddy, rocky and vegetated coasts. The number of used training transects is on the x-axis and the accuracy (f1-score) on the y-axis. The solid line represents the mean accuracy, the shaded area represents the 95th percentile of the accuracy. Right-hand panel b: importance of features (calculated as the mean decrease in impurity method) used in the hybrid transect classifier in descending order. The bars represent the mean of the importance for each feature originating from the multispectral satellite images (blue) and the coastal geospatial datasets (orange). The black lines represent the standard deviation of the importance of each feature.