Land cover classification using high‐resolution aerial photography in adventdalen, svalbard

Abstract

A methodology was tested for high‐resolution mapping of vegetation and detailed geoecological patterns in the Arctic Tundra, based on aerial imagery from an unmanned aerial vehicle (visible wavelength – RGB, 6 cm pixel resolution) and from an aircraft (visible and near infrared, 20 cm pixel resolution). The scenes were fused at 10 and 20 cm to evaluate their applicability for vegetation mapping in an alluvial fan in Adventdalen, Svalbard. Ground‐truthing was used to create training and accuracy evaluation sets. Supervised classification tests were conducted with different band sets, including the original and derived ones, such as NDVI and principal component analysis bands. The fusion of all original bands at 10 cm resolution provided the best accuracies. The best classifier was systematically the maximum neighbourhood algorithm, with overall accuracies up to 84%. Mapped vegetation patterns reflect geoecological conditioning factors. The main limitation in the classification was differentiating between the classes graminea, moss and Salix, and moss, graminea and Salix, which showed spectral signature mixing. Silty‐clay surfaces are probably overestimated in the south part of the study area due to microscale shadowing effects. The results distinguished vegetation zones according to a general gradient of ecological limiting factors and show that VIS+NIR high‐resolution imagery are excellent tools for identifying the main vegetation groups within the lowland fan study site of Adventdalen, but do not allow for detailed discrimination between species.

Comparing ultra‐high spatial resolution remote‐sensing methods in mapping peatland vegetation

“…Spatial patterns of peatlands can be tracked with remotely sensed data, but it has been argued that the spatial resolution in common mapping approaches is too coarse (Palace et al, 2018). Ultra-high spatial resolution (UHSR) remote sensing, which provides data with cm-level pixel size, can reveal such patterns in vegetation composition that are lost in coarser resolution (Díaz-Varela, Calvo Iglesias, Cillero Castro, & Díaz Varela, 2018;Gonçalves et al, 2016;Lehmann et al, 2016;Mora, Vieira, Pina, Lousada, & Christiansen, 2015). In particular, the benefits of UHSR are evident in fragmented landscapes such as peatlands (Arroyo- Mora, Kalacska, Lucanus, Soffer, & Leblanc, 2017;Lehmann et al, 2016;Lovitt et al, 2017;Palace et al, 2018).…”

UAV photogrammetry for mapping vegetation in the low-Arctic

“…In one case, a fixed-wing UAV was used to acquire visible and infrared imagery to assess the separability of a range of Arctic plant communities in Svalbard based on RGB and NDVI images (Tømmervik et al 2014). In a nearby study area, Mora et al (2015) fused UAV RGB imagery with visible and NIR imagery from manned aircraft to classify major vegetation types using four supervised classification methods. Both studies used spectral information only and did not examine the potential utility of height or structural information derived from SfM point clouds.…”

Species diversity of forest stands estimation algorithm using RGB images of the tree crowns

“…Zakładamy zatem, że do dalszych analiz dysponujemy zobrazowaniem w trzech szerokich kanałach optycznych promieniowania widzialnego o wysokiej efektywności kwantowej i rozdzielczości co najmniej kilku megapikseli. W przeciwieństwie do profesjonalnych badań zróżnicowania pokrycia terenu za pomocą zdjęć satelitarnych [16] i lotniczych [20] za pomocą samolotów bezzałogowych [11] oraz lekkich samolotów załogowych [17,31] zobrazowania wykorzystywane w tej pracy wykonywane były z poziomu terenu, a nie z pułapu rzędu kilkuset metrów (zdjęcia lotnicze) lub kilkuset kilometrów (satelitarne). Do danych obrazowych tego typu nie można zatem wprost zastosować typowych algorytmów oszacowania różno-rodności używanych w klasycznym postprocessingu.…”