A Post-Classification Change Detection Model with Confidences in High Resolution Multi-Date sUAS Imagery in Coastal South Carolina
- 2 March 2021
- journal article
- research article
- Published by Taylor & Francis in International Journal of Remote Sensing
- Vol. 42 (11) , 4309-4336
Abstract
Small unmanned aerial systems, or sUAS, remote sensing has much potential for monitoring vegetation cover and other phenomena due to its innate ability to capture very high spatial resolution imagery at low altitudes, for small areas, and with rapid planning. Remote sensing methodologies used in monitoring change from orbital and manned aerial remote sensing have been extensively researched and implemented. However, these same historic change detection methodologies have not been thoroughly tested with the new, very high spatial resolution imagery captured by sUAS, particularly the modest sUAS proliferating in the resource management agencies around the world. This study seeks to examine and understand the variability involved in using sUAS for change detection of vegetation cover by designing a confidence model, calibrating the model, and demonstrating the use of the model. Our research design involves a novel approach using multiple collections in a 1.6-hour period over a controlled environment where the land cover did not change. A confidence model was developed and calibrated in the controlled environment. A demonstration of the developed confidence model from the control environment was used for a hurricane impacted area. Coastal dune vegetation cover, essential for dune strength and growth, was monitored before and after Hurricane Irma impacted Harbor Island in coastal South Carolina. The results indicate that even though no actual change occurred during the controlled experiment, an average of 5.6% of the pixels indicated a false change of land cover. These false land cover discrepancies are caused from slight shadow movements, georegistration accuracies, multiple look angles and variable spectral response from a mosaic (i.e. 120 images) of imagery. It was also determined that much change in vegetation cover occurred as a result of inundation from Hurricane Irma, and confidences were high in assessing the change.Keywords
This publication has 42 references indexed in Scilit:
- Development of a UAV-LiDAR System with Application to Forest InventoryRemote Sensing, 2012
- Small-Scale Unmanned Aerial Vehicles in Environmental Remote Sensing: Challenges and OpportunitiesGIScience & Remote Sensing, 2011
- Central Limit Theorem2010
- Remote sensing change detection tools for natural resource managers: Understanding concepts and tradeoffs in the design of landscape monitoring projectsRemote Sensing of Environment, 2009
- Extending post-classification change detection using semantic similarity metrics to overcome class heterogeneity: A study of 1992 and 2001 U.S. National Land Cover Database changesRemote Sensing of Environment, 2008
- Spatial Modeling and Analysis for Shoreline Change Detection and Coastal Erosion MonitoringMarine Geodesy, 2001
- Monitoring land-cover changes: A comparison of change detection techniquesInternational Journal of Remote Sensing, 1999
- The use of the empirical line method to calibrate remotely sensed data to reflectanceInternational Journal of Remote Sensing, 1999
- Review Article Digital change detection techniques using remotely-sensed dataInternational Journal of Remote Sensing, 1989
- Radiometric scene normalization using pseudoinvariant featuresRemote Sensing of Environment, 1988