Remote sensing for monitoring restoration efforts in Alpine areas

The global loss of biodiversity is one of our biggest current challenges. A quantitative and objective monitoring system assisted by remote sensing is needed especially for inaccessible mountain regions in order to identify areas of restoration needs and also to monitor the impact of restoration efforts. Mountain regions like the Alps are known biodiversity hotspots, but, at the same time, the rough topography and harsh climate make frequent and walltowall assessments in the field difficult and expensive. Earth Observation (EO) techniques have been applied to monitor land cover, vegetation, and other environmental factors over time in an efficient manner, but their use for biodiversity assessment and restoration monitoring is not yet fully exploited. Which EO data and methods are useful to provide walltowall data on restoration needs and success? How well can indicators derived from various EO data sets be used to monitor impacts of the Nature Restoration Law? Highresolution LiDAR data is used to determine the restoration of previously disturbed forest areas using foliage height diversity (FHD) as an indicator. For grassland monitoring, we use optical satellite time series images and insitu data to derive indicators like mowing intensity and grassland types. Existing data on structural landscape features in agricultural areas is used to compare regions and identify areas with a high restoration need. FHD is a useful indicator to monitor forest restoration after disturbance events like bark beetle outbreaks. Instead of only evaluating the reclosure of canopy cover, FHD also integrates the vertical structure, thus giving a much more realistic assessment. Tests in an area heavily affected by bark beetles revealed that, based on the FHD, almost 65% of the area was not yet fully restored. For agricultural areas, we used data from two sources: Copernicus Small Woody Features (SWF) layer and a Habitat Type Assessment (HTA) done some 10 years ago. Our assessments found that highest detection rates (62% compared to 45% HTA and 31% SWF respectively) were achieved when combining the data sets because of the complementarity for many of the structural features. For example, 26% of all hedges were detected both by SWF and HTA, but combining them led to a detection rate of almost 44%.