Hyperspectral modeling of ecological indicators – A new approach for monitoring former military training areas

Abstract

Military areas are valuable habitats and refuges for rare and endangered plants and animals. We developed a new approach applying innovative methods of hyperspectral remote sensing to bridge the existing gap between remote sensing technology and the demands of the nature conservation community. Remote sensing has already proven to be a valuable monitoring instrument. However, the approaches lack the consideration of the demands of applied nature conservation which includes the legal demands of the EU Habitat Directive. Following the idea of the Vital Signs Monitoring in the USA, we identified a subset of the highest priority monitoring indicators for our study area. We analyzed continuous spectral response curves and tested the measurability of N=19 indicators on the basis of complexity levels aggregated from extensive vegetation assemblages. The spectral differentiability for the floristic as well as faunistic indicators revealed values up to 100% accuracy. We point out difficulties when it comes to distinguishing faunistic habitat requirements of several species adapted to dry open landscapes, which in this case results in Overall accuracy of 67, 87–95, and 35% in the error matrix. In summary, we provide an applicable and feasible method to facilitating monitoring military areas by hyperspectral remote sensing in the following.