Analysis of riparian forest habitats on the basis of very high resolution satellite imagery – a test and transfer exercise within the MS.MONINA ‘site-level service’
Riparian forests play a significant role in safeguarding biodiversity and related ecosystems services. Despite their significance as ‘biodiversity hot spots’ little is known in terms of their actual and likely changing conditions under a series of pressures from outside (surrounding infrastructures, anthropogenic activities including forestry, hunting, leisure as well as climate change). Within Europe, the prevailing multi-policy framework (comprising the Habitats Directive, the Water Framework Directive, the Flood Directive, etc.) entails strategies for the protection of these sensitive areas. As the area covered by riparian zones is rather small (approx. 2% of the European land surface), we need to apply fine-scaled methods and data when assessing the integrity of these ‘lifelines’.
The use of Earth observation (EO) has matured over the last years, both in terms of analytical power and the provision of details which are necessary for fine-scaled habitat assessments The so called ‘site-level service’, developed within MS.MONINA, aims at providing detailed EO-based information products on the status, quality, change and pressure of specific ecosystems. One crucial step is the generation of a robust and transferable habitat mapping scheme, which utilizes the intrinsic information power of latest EO-data (e.g. 8-band WorldView-II imagery) and captures the specificities of habitat arrangements and internal structures.
In the study presented, we performed a transfer exercise from one riparian forest ecosystem to a comparable one, based on the concept of an implemented object-feature library (OFL) and a transfer protocol. The idea is to optimize the effort required for collecting field reference data, which is necessary for both verification (during the classification process) and validation (to test the accuracy of the habitat map). Habitats were delineated using a class modeling approach resulting in a multi-scale representation of the scene content on three levels: tree stands, elementary forest type units and habitats according to Annex I of the Habitats Directive. Based on the three validated hierarchical layers, a quality assessment for EO-data derived information was composed by using a set of indices in terms of composition and conservation status.