Habitat sampler—A sampling algorithm for habitat type delineation in remote sensing imagery

Abstract

AbstractAimThe management of habitats for the conservation and restoration of biodiversity in protected area networks requires an appropriate monitoring to increase our understanding of processes and dynamics in managed ecosystems. Remote sensing offers a unique potential for the derivation of coherent spatiotemporal information to report on natural or management‐induced habitat change. However, the methods used for the delineation of habitat types in remote sensing imagery depend on the extensive process of ground truth sampling as reference to construct image classifiers. In fact, the number of required reference samples is intrinsically unknown in complex scenes due to the heterogeneity of varying habitat conditions. Thus, most classifiers are not transferable in retrospective image analysis or between different ecosystems that is, however, required for an operational use of remote sensing‐based monitoring systems.InnovationA new procedure is introduced that autonomously generates representative reference samples for a predictive modelling of habitat type probabilities. The procedure, termed Habitat Sampler, is provided as a tool that can be applied to any image input that display vegetation structures and dynamics on multiple temporal and spatial scales. The Habitat Sampler provides many labelled point locations for the training of image classifiers and enables a fast and easy to implement model transfer for the delineation of habitat dynamics in various ecosystems.Main conclusionsThe Habitat Sampler outperforms standard machine learning classifiers when the distribution of reference samples is unknown or insufficient. It was shown that particularly in retrospective image analyses patterns of successional and cyclic habitat development can be mapped for large heathland areas. The procedure is feasible for application in biodiversity conservation monitoring using various habitat typologies that are associated over ecosystem processes, particularly to report on protected area networks using cost‐free satellite imagery.