Coastal Restoration Trust of New Zealand

Coastal Dune Ecosystem Reference Database

Unmanned Aerial System derived Multi-Spectral Imagery for the Monitoring of Coastal Dune Plant Communities Thesis

Author
Fake, Michael
Year
2019
Publisher / Organisation
Lincoln University
Pages
118
Species
Ammophila arenaria, marram, marram grass, Spinifex sericeus, kowhangatara, spinifex, silvery sand grass, Lupinus arboreus, Ficinia spiralis, pingao, pikio, golden sand sedge, Desmoschoenus spiralis, Muehlenbeckia astonii, shrubby tororaro, wiggywig, mingimingi, Carex pumila, sand sedge, dune sedge, Calystegia soldanella, rauparaha, nihinihi, shore bindweed, shore convolvulus, Poa billardierei, hinarepe, sand tussock, Austrofestuca littoralis, Poa cita, silver tussock, Pimelea prostrata, pinatoro, native daphne, Carmichaelia appressa, prostrate broom, Craspedia "Kaitorete", Lycium ferocissimum, boxthorn, Melicytus alpinus, porcupine shrub, Pinus radiata, radiata pine
Keywords
Remote Sensing, UAS, UAV, Drone, Image Classification, Plant Community, Vegetation, Coastal, Sand Dunes, Ordination, TWINSPAN, Clustering
Summary
Plant community monitoring was conducted at Kaitorete Spit Scientific Reserve using UAS based remote sensing and traditional field-based techniques. Multi-spectral, high resolution UAS imagery was used as the basis for image classification. Different classification methods and data manipulation techniques were evaluated in order to present the most accurate representation of plant communities for comparison against those derived from the field data. Overall image classification results were on par with those from similar studies, however their suitability for application to the monitoring of the specific environmental and ecological conditions at Kaitorete Spit remains of low confidence. UAS imagery was able to be used to identify coarse scale ecological features which could then be used to define distinct ecological communities in a similar but not identical manner to that of the field data. At a finer-scale, UAS imagery could detect some, but not all, key ecological features such as individual species or fine-scale indicators of diverse habitat types.