ABSTRACT
Unmanned aerial vehicles (UAVs) have the potential to provide highly detailed information on vegetation status useful in precision agriculture. However, challenges are associated with existing techniques for UAV-based retrieval of vegetation biophysical variables such as leaf area index (LAI), including variable illumination, bidirectional reflectance effects, and the need for image calibration, mosaicking, and normalization. We investigated an alternative approach that avoids these challenges whilst still providing spatially explicit estimates of LAI, using UAV-based digital hemispherical photography (DHP). LAI estimates were obtained using a low-cost UAV-based DHP system over a winter wheat field in Southern England. Point-based estimates were interpolated to provide spatially continuous datasets, which successfully described patterns of vegetation condition. The UAV-based DHP data were compared to ground-based LAI estimates, demonstrating good agreement (root mean square error (RMSE) = 0.10, normalized RMSE (NRMSE) = 3%).
Acknowledgments
The authors are grateful to Lord Camden, Gwyn MacFarlane and Steve Shean for facilitating access to Wherwell Farms, and also thank Peter Morgan and the University of Southampton School of Geography and Environmental Science laboratories team for field assistance.