Airborne Scanning

András Zlinszky et al: Mapping Natura 2000 habitat conservation status in a Pannonic Salt Steppe with Airborne Laser Scanning 20.03.2015

Despite their importance for biodiversity, grasslands are rarely monitored with remote sensing, and hardly ever with airborne laser scanning. The differences between individual grassland vegetation classes in terms of height and texture are very small compared to forests or shrublands, while the necessary spatial resolution is much higher. An international team led by Prof. Norbert Pfeifer of TU Vienna and dr. András Zlinszky of the Hungarian Centre for Ecological Research processed point clouds acquired from alkali grasslands in Püspökladány, Hungary, mapping not only vegetation categories but the conservation status of the habitat. Both leaf-on and leaf-off datasets were acquired with a RIEGL LMS-Q680i scanner. Intensity information was calibrated to reflectance based on field spectrometer measurements. A multi-parameter dataset of reflectance, echo width, canopy height and various texture indicators was processed using purpose-built machine learning classification software. The result was a series of vegetation maps with a resolution of 0.5 m and up to 15 categories of herbaceous vegetation. These maps were further analysed in a GIS algorithm that followed the guidelines of Natura 2000 conservation status evaluation: 12 out of 13 ecological variables requested by this scheme were successfully calculated from the point cloud. These data layers were finally evaluated to deliver a map of habitat quality of the grassland, which was proven to be 80% correct compared to 10 independent field samples. This level of reliability is similar to the agreement expected between two independent field measurements.

The study shows that airborne laser scanning has sufficient information content to serve as a standalone data source for habitat monitoring even in such a highly complex vegetation mosaic.

The full article was published Open Access in the journal Remote Sensing, 2015, doi:10.3390/rs70302991 and can be found here