UAV-based Imaging of Multispectral Reflectance and Solar-Induced Chlorophyll Fluorescence for Crop Monitoring

Abstract

The agri-food sector is facing significant challenges due to climate change, unpredictable weather, and rapid population growth. The growing demand to embrace advanced agricultural systems that boost productivity while mitigating climate impacts requires accurate and reliable crop monitoring techniques. In this context, site-specific farm management and remote sensing have become indispensable. Remote sensing offers real-time information about crop growth and health throughout the growing season at different scales. UAV-based remote sensing, in particular, offers a cost-effective tool for monitoring crop growth and health with high spatiotemporal resolution that enables response to field-scale issues by driving informed decision-making. Our study contributes to this evolving landscape by exploring the potential of UAV-based multispectral imaging in crop monitoring on the field scale. <br /> In the first study, high-resolution imagery from the DJI Phantom 4 multispectral UAV system was employed to monitor the seasonal development of spelt in a biochar-enriched experiment. A straightforward data processing workflow was developed based on an empirical line method to convert raw UAV data to normalized and comparable reflectance maps and calculate various vegetation indices. Results indicated that EVI was the most effective index in relation to the actual yield, indicating better spelt development over the biochar-enriched plots with a full conventional fertilization amount compared to controls that received the same conventional fertilization. <br /> The second study addressed the retrieval of sun-induced fluorescence, F760, from SIFcam, a dual-camera system prototype mounted on a UAV. A comprehensive overview and advancements of the developed methodology for SIFcam imagery, in addition to a second innovative workflow, were presented. The F760 retrieved from the two workflows showed strong correlations with ground-based measurements (R² = 0.92) and moderate correlations with airborne imaging spectrometer HyPlant (R² = 0.56, 0.52 for workflows 1 and 2b, respectively). The SIFcam has shown its capability to effectively disentangle the fluorescence signal from canopy reflectance with a moderate level of accuracy and adequate stability in data collection at the field scale, with less than one-pixel variation between spectral channels in both horizontal and vertical directions. <br /> The third study investigated the potential of integrating SIFcam F760 alongside UAV-based multispectral VIs to characterize and delineate diverse new and old winter wheat cultivars. SIFcam demonstrated a notable potential in capturing the variability of F760 between wheat cultivars with structural and pigment differences. New wheat cultivars generally revealed higher F760, consistent with their higher chlorophyll content, yet old cultivar Banco indicated that canopy architecture could significantly modulate TOC F760, with F760 values comparable to or even exceeding those of certain new cultivars. VIs sensitive to chlorophyll content, particularly TCARI/OSAVI (Cohen's d >= 0.5), outperformed structure-related VIs and F760 for distinguishing the cultivars. SIFcam proved to be a valuable tool for field plant phenotyping and potentially guiding breeding programs.