Evaluation of high-resolution imagery for monitoring the flowering of holm oak trees

The Mediterranean oak savanna is Europe's most extensive agroforestry system, with significant economic, social, and ecological relevance. Climate models indicate that the Mediterranean region is particularly vulnerable to the impacts of global warming, which include increased frequency and severity of droughts. Consequently, there is a pressing need for conservation measures to prevent the degradation of this ecosystem, reduce uncertainty about production, and ensure its sustainable development. One of the most valuable resources of this system is the acorn of holm oaks, which significantly contributes to the quality of extensive livestock products. The intensity of oak flowering is a key factor limiting maximum acorn production, and it is usually monitored by visual sampling in the field, a costly and time-consuming method. To address this challenge, this study aims to evaluate the potential of remote sensors onboard UAVs (unmanned aerial vehicles) and high-resolution satellite sensors for monitoring the flowering of holm oak trees. This approach aims to scale up the monitoring effort, simultaneously providing valuable information to many farmers.

 

Previous works have explored the use of digital cameras for monitoring phenology, particularly in combination with airborne data. Gómez-Giráldez et al. (2021) proposed an index designed to automatically quantify the male flowering intensity of holm oaks based on the closeness to pure yellow in RGB images captured by UAVs. This index showed sufficient accuracy in differentiating between various flowering intensity levels and providing intensity maps. However, the index requires further validation before it can be applied on a larger scale. 

 

This study extends the validation of the proposed index using UAV data and high-resolution satellite imagery. It was conducted on six plots located in southern Spain. During 2022 and 2024, 11 images were taken over the three previously studied plots and three additional plots to enhance the validation. The results indicated that trees with lower flowering intensity were concentrated in areas with higher yellow distances, which confirmed previous results. However, due to the high phenological variability among individuals, we observed the importance of synchronizing the image acquisition date with the peak flowering period. Two high-quality orthoimages with a resolution of 1.5 m were acquired from the SPOT 6 satellite in 2024 to extend the methodology to larger areas and provide intensity maps. 

 

The expanded evaluation and visual verification showed promising results. The index, which can be derived using just an RGB image, shows potential for future applications related to phenology and productivity. Furthermore, developing an automated tool for this task would be beneficial in covering large areas and improving the representativeness of the estimates.