Synergetic use of Sentinel‐1 and Sentinel‐2 for assessments of heathland conservation status

Abstract

Habitat quality assessments often demand wall-to-wall information about the state of vegetation. Remote sensing can provide this information by capturing optical and structural attributes of plant communities. Although active and passive remote sensing approaches are considered as complementary techniques, they have been rarely combined for conservation mapping. Here, we combined spaceborne multispectral Sentinel-2 and Sentinel-1 SAR data for a remote sensing-based habitat quality assessment of dwarf shrub heathland, which was inspired by nature conservation field guidelines. Therefore, three earlier proposed quality layers representing (1) the coverage of the key dwarf shrub species, (2) stand structural diversity and (3) an index reflecting co-occurring vegetation were mapped via linking in situ data and remote sensing imagery. These layers were combined in an RGB-representation depicting varying stand attributes, which afterwards allowed for a rule-based derivation of pixel-wise habitat quality classes. The links between field observations and remote sensing data reached correlations between 0.70 and 0.94 for modeling the single quality layers. The spatial patterns shown in the quality layers and the map of discrete quality classes were in line with the field observations. The remote sensingbased mapping of heathland conservation status showed an overall agreement of 76% with field data. Transferring the approach in time (applying a second set of Sentinel-1 and -2 data) caused a decrease in accuracy to 73%. Our findings suggest that Sentinel-1 SAR contains information about vegetation structure that is complimentary to optical data and therefore relevant for nature conservation. While we think that rule-based approaches for quality assessments offer the possibility for gaining acceptance in both communities applied conservation and remote sensing, there is still need for developing more robust and transferable methods.

Climatic and vegetational drivers of insect beta diversity at the continental scale

“…The open data policy of Landsat and Sentinel potentially allows for such an approach, and recently published datasets aiming at capturing forest cover dynamics and human disturbance (Hansen et al, ) provide starting points to explore this approach. Further, the free, global availability of Sentinel‐1 SAR data potentially enable the assessment and incorporation of vegetation structural attributes (Schmidt et al, ) into large‐scale assessments and analyses of insect communities. Though there have been several promising studies that showed correlations between remote sensing products and ecosystem structure, habitat conditions, and animal communities, it must be acknowledged that we are still in an early stage of deriving reliable indicators for biodiversity information from earth observation data (Bush et al, ).…”

Mapping grassland plant communities using a fuzzy approach to address floristic and spectral uncertainty

“…Until now, accuracy greater than 85% could be reached only by using airborne hyperspectral (Burai, Deák, Valkó, & Tomor, ), hyperspatial resolution imagery from an unmanned aerial vehicle (Kaneko & Nohara, ) or full‐waveform LiDAR data (Launeau et al., ) of small sites covering several hectares. However, the new Sentinel time‐series with a high spatial resolution (10 m) seems promising to accurately map natural habitats, using Sentinel‐1 SAR (Schmidt, Fassnacht, Förster, & Schmidtlein, ) or Sentinel‐2 multispectral (Shoko & Mutanga, ) sensors. Such time‐series satellite images, which are both free and cost‐effective, are now available and appear promising to accurately distinguish between plant communities in the near future.…”

“…Here, we used RS data to map nine “elementary” habitats that are included in the “thermo‐Atlantic and sub‐brackish meadows” Natura 2000 habitat, which encompasses diverse and even contrary environmental conditions caused by environmental management such as flood duration (from less than 1 month for Hordeo secalini‐Lolietum perennis up to 4 months for Eleocharito palustris‐Oenanthetum fistulosae ) and agricultural practices (grazing for Carici divisae‐Lolietum perennis or mowing for Junco gerardi‐Oenanthetum fistulosae ). From a structural perspective, each “elementary” habitat that we mapped could be used as a surrogate for a species pool (Zlinszky et al., ) but also for species richness, stand structural diversity or key species cover (Neumann et al., ; Schmidt et al., ). From a functional perspective, these “elementary” habitats—which are ecologically homogeneous units—could also provide insights into their corresponding environmental conditions as well as their functional traits (e.g., seed mass), which are indirectly detectable from RS data (Violle et al., ).…”

Using Airborne Hyperspectral Imaging Spectroscopy to Accurately Monitor Invasive and Expansive Herb Plants: Limitations and Requirements of the Method

“…Depending on the area, the training polygons occupied from 0.01% to 0.07% of the surveyed area (calculated area covered by the target species polygons). It can therefore be assumed that a set of polygons, optimized in terms of the percentage cover of a species, allows for a correctly assessed classification with a significantly lower sampling rate compared to 1–2% of the area described in the literature [59]. The correct result of identification on a comparable set of training polygons was achieved with the use of the Mixture Tuned Matched Filtering (MTMF) algorithm [58] when mapping Cardaria draba .…”