Authors: Grayson R Morgan*, University of South Carolina, Michael E Hodgson, University of South Carolina
Topics: Remote Sensing
Keywords: Drones, sUAS, Coastal, MultiSpectral, Change Detection, Remote Sensing
Session Type: Paper
Start / End Time: 1:10 PM / 2:50 PM
Room: Harding, Marriott, Mezzanine Level
Presentation File: No File Uploaded
Small unmanned aerial systems (sUAS)-based image collection has made available affordable and repetitive monitoring of coastal marsh vegetation. However, unlike satellite or airborne imagery, personal sUAS collect imagery at much lower altitudes and thus, the imagery is of higher spatial resolution. As a result, the imagery is at oblique angles through much of the image. Little is known about the sensitivity of drone acquired imagery at such resolutions and viewing angles for monitoring changes in a landscape. In this study we propose an experimental design to examine the sensitivity of drone imagery to repeat image collections at the 'same' photo centers at low altitudes. We defined confidence intervals for apparent (but not actual) changes in vegetation caused simply by small shifts in photo centers typical of repeat image collections. Spectral differences on ‘same’ flight paths are quantified by comparing 15 spectral ground controls in the flight path among subsequent flights. Data gathering sorties will be flown at 12:00 and then 12:15 to ensure optimal conditions among flights along the same flight paths. Analysis is conducted using ground control and without ground control. Following data gathering and processing, statistical significance of spectral differences between ‘same’ flight path’s spectral information will be calculated using a Student T test. We conclude the analysis by offering thresholds for defining an actual versus ‘apparent’ vegetation or land cover change.