Hyperspectral remote sensing for ecological analyses of grassland ecosystems - spectral separability and derivation of NPP-related biophysical and biochemical parameters

This study has the overall objective to develop, apply, and test a tool based on modern GIS and remote sensing techniques and data suitable for identification, planning and monitoring in nature conservation (specifically in the framework of the TWW project). Tests will be performed to evaluate the applicability of the technique for selected TWW monitoring questions. .

We will analyze the historical range of variability and thresholds (of VI’s, LAI, NPP) from multi-temporal remotely sensed imagery for mapped TWW polygons composited for individual months throughout the year. This historical range of variability and its relation to climate pattern will then be used to derive thresholds of change/no-change for monitoring the efficiency of the protective actions for maintaining these habitats. Such threshold will allow us to identify TWW polygons that show unusual deviations from the expected historical signal, thus suggesting that the vegetation inside of this polygon has changed unexpectedly. The method is applicable to other ecosystems as well, but we will specifically use TWW data to develop and test the method.

In a second application, we suggest using the available mapped TWW polygons to train a multi-temporal regression model to identify "missing polygons" using GLM (Generalized Linear Models). This approach will allow us to find pixels that show essentially the same historical signal as pixels that have been mapped and identified. This can be valuable to assist the TWW teams in areas where the a priori knowledge of the distribution of dry meadows and pastures is suboptimal. Additionally, testing such an approach will help evaluating this method for later possible additions to the acknowledged TWW polygons once the mapping project has been finished.