Most informative spectral indices in terms of discriminatory power for cluster differentiation

The supplementary materials provide an expanded methodological account of unsupervised classification approaches for monitoring rapid post-disaster transformations of landscape cover, together with the outputs of their application to the floodplain ecosystems of Khortytsia Island (Dnipro River, Ukraine), which underwent major change following the 2023 Kakhovka dam failure and the subsequent drawdown of the Kakhovka Reservoir. Particular emphasis is placed on how, under highly dynamic and structurally uncertain conditions of post-catastrophe succession, combining ordination procedures (including sequential principal component analysis) with hierarchical clustering enables the delineation of both dominant surface types and subtler, ecologically meaningful differences related to moisture conditions, soil/substrate properties, and vegetation status. The materials also summarise the discriminatory (information) value of spectral indices for cluster differentiation and illustrate seasonal shifts in the leading spectral signals across the growing season. The mapped results document spatiotemporal differentiation of land-cover classes, the emergence of new and transitional surface types, and their trajectories of change throughout 2023. Land-surface classification outputs for all analysed dates are provided in our dataset.