Authors: Chunhua Zhang*, Algoma University, John M. Kovacs, Nipissing University, Dan Walters, Nipissing University
Topics: Remote Sensing, Geographic Information Science and Systems, Agricultural Geography
Keywords: unmanned aerial system, precision agriculture, high resolution satellite imagery, WorldView-2, crop condition
Session Type: Paper
Start / End Time: 10:00 AM / 11:40 AM
Room: Balcony M, Marriott, River Tower Elevators, 4th Floor
Presentation File: No File Uploaded
The rapid development of small unmanned aerial systems (UAS) in recent years has triggered a considerable interest in their application for precision agriculture. The decreasing cost, the ultra-high spatial resolution, and the increased flexibility of image acquisition have made UAS remote sensing an ideal tool for monitoring crop conditions. Working in collaboration with farmers in Verner, Ontario, we examined the efficacy of near-infrared UAS imagery in monitor crop conditions, in comparison to high resolution satellite imagery. Crop biological data were collected every two weeks during the growing season. UAS images were captured using an ADC lite camera onboard an Aeryon Scout UAS approximately the same time of field data collection. One WorldView-2 image was acquired on August 8, 2014 for comparison purpose. UAS images were calibrated and orthorectified to create several mosaic images of the study sites. ATCOR correction was applied to the WorldView image to remove the atmospheric impacts. Digital numbers (DNs) and reflectance of the sampling plots were extracted from the UAS image mosaics and the WorldView image based on their GPS coordinates and imagery interpretation. Various vegetation indices were also calculated based on the DNs. Pearson’s correlation and linear regression analyses were then used to examine the relationships between the DNs and the vegetation indices with the corresponding crop canopy biophysical parameters. The results indicate that there are strong correlations between crop biological data and the image reflectance created from UAS and satellite imagery and, moreover, that these relationships can be successfully modelled using linear regression techniques.