Midwest Fish & Wildlife Conference 2017

AUTHORS: 1) Deon van der Merwe, Associate Professor, Kansas State University (KSU), Department of Diagnostic Medicine/Pathobiology; 2) Ajay Sharda, Assistant Professor, KSU, Department of Biological & Agricultural Engineering; 3) Dale Bremer, Professor, KSU, Department of Horticulture & Natural Resources; 4) Lee R. Skabelund, Associate Professor, KSU, Department of Landscape Architecture and Regional & Community Planning

ABSTRACT: Diverse, living vegetative coverage is deemed vital to optimize ecosystem services on green roofs. Kansas State University (KSU) faculty and students initiated studies of vegetation growing on two large new green roofs on the KSU campus. Combined, the Memorial Stadium green roofs (MS-GRs) were seeded and planted with 30 native species and encompass more than 34,000 square feet. The efficiency and dimensionality of vegetation assessment at this scale may be effectively enhanced by utilization of aerial sensors to provide measurements that are correlated with vegetation cover and growth vigor.
Thermal (FLIR VUE PRO 640) and color-infrared (modified Canon S100) sensors were used on a small unmanned aircraft system (sUAS; 3DR Iris+). Due to the relatively small payload capacity of the sUAS, the two sensors were used sequentially on separate flights. The flights were completed within an hour of noon on a clear day in early July 2016. Images were collected from an altitude of 50m, with 80% overlap front to back and side to side. The images were used to construct surface elevation models (SEMs) and orthomosaics using a surface invariant feature transform method (Agisoft-Photoscan-Pro), where the reflectance values were averaged from all contributing images, and the models were georeferenced based on a series of ground control points. The thermal orthomosaics were calibrated based on a series of ground targets with known surface temperatures, and transformed into temperature maps. The color-infrared orthomosaics were transformed into normalized difference vegetation index (NDVI) maps.
Results indicated that green roof surface temperature and NDVI values were correlated with vegetative cover. Data collection was highly efficient, with each flight completed in less than five minutes. We concluded that sUAS-based sensors provide a viable and efficient data collection approach, with practical application in vegetation characterization on green roof installations.