UAV-based Imaging for Multi-Temporal, very high Resolution Crop Surface Models to monitor Crop Growth Variability

This paper describes the generation of multi-temporal crop surface models (CSMs) with very high resolution of < 0.05 m. Data collection was carried out with a low-cost and low-weight UAV-system with a weight of less than 5 kg and the possibility of mounting different sensors. Key focus is the detection of crop growth variability and its dependency on cultivar, crop treatment and stress. The study area is a barley experiment field in Bonn in the west of Germany. Four replications of four cultivars of barley were investigated of which half of them where treated with a fungicide. Five UAV-campaigns were carried out during the growing season between early May and late July 2012. Ground control points (GCPs) measured with a Hi-Per Pro Topcon DGPS allowed for appropriate ground truth (< 0.02 m). Ground based infield control surveys on three dates served as validation of the method. Additionally, various destructive and non-destructive ground data were collected. The stereo images captured were processed into CSMs by using the structure-from-motion (SfM) software Agisoft PhotoScan. Generated plant heights ranged between 0.16 m and 0.983 m. R-2 (n = 32) for the correlation between plant heights in the CSM and infield control surveys is 0.69. Lower plant heights were detected in those plots of the field where no fungicide was applied. Height differences between cultivars were observed and increased during growing season. The accuracy assessment of DEMs generated with the proposed UAV-based imaging showed a correlation coefficient of 0.99 (n = 10) between the DGPS GCPs and the DEMs with a mean difference of 0.01 m in z-direction.