Scaling SIF: from point observations to the entire field using UAV-based SIF and multispectral data

By Dong Liang

Abstract
The increasing global population has resulted in insufficient capacity to feed everyone. Thus, enhancing crop yields is crucial to guaranteeing long-term global food security. Solar-induced fluorescence (SIF) is an important tool for monitoring photosynthesis, a core part of the physiological process in plants that determines the maximum achievable yield of crops. The UAV-based FluorSpec system offers a promising approach to bridging point-based SIF measurements with field-scale SIF mapping, enhancing our ability to assess crop performance and stress responses. This study aims to identify a robust method for scaling SIF from point observations to the entire field using vegetation indices and plant traits derived from multispectral data and to evaluate the importance of these predictive features in predicting SIF. This research found that combining ATP kriging and the prediction of the random forest model is the most robust method. In the early growth stage of the crop, soil-resistant vegetation indices like GEMI, and structural vegetation indices contributed the most to the SIF prediction. When the crops were in the middle growth stage, the chlorophyll-related, structural and red-edge-based vegetation indices were the dominant inputs in the SIF prediction. In the mature stage before harvest, the red-edge-based vegetation indices are the most important in predicting the SIF of crops. This research provides a robust and innovative method for generating high-resolution field-scale SIF maps using the most affordable UAV-mounted sensors. It offers a cost-effective solution for monitoring crop photosynthesis in precision agriculture, providing data support to improve crop management, optimise yields, and support sustainable agricultural practices.