Remote sensing for the observation of senescence in Conference pear trees

Leaf senescence in trees is the phenological stage during which nutrient resorption happens. In this process, part of the nutrients is transferred to the perennial organs of the plant, contributing to tree vitality and, in pome fruit trees, to flowering intensity the following year. Another share of the nutrients remains inside leaf litter and enters the agroecosystem’s nutrient cycles. The timing and duration of leaf senescence influences the ratio between the two parts of nutrients and thus influences nutrient cycles in the agroecosystem. Among innovative techniques to investigate these processes, satellite remote sensing has proved a valid tool in natural ecosystems. The same cannot be said about fruit orchards, because of the image quality of the satellites active before Sentinel-2, often deemed insufficient for agricultural studies. The features of Sentinel-2, instead, offer new possibilities for monitoring phenology in agricultural environments. This research aims to study senescence in Conference pear trees, in three regions of Flanders (Belgium). One cloud-free Sentinel-2 image, acquired in the middle of the senescence period, was analysed, by means of different spectral indices. Ground data was collected through a network of 34 webcams with an RGB camera. A visual analysis was performed, to determine the beginning of senescence (the moment in which the first yellow/red leaves appear in the canopy) and the end of senescence (the moment in which the entire canopy turns yellow/red). Webcam data showed that leaf (dis)colouration started between September and October, during a one-month time span. Full discolouration of the canopy, occurring at the end of November, was instead more synchronous. Moreover, some trees only turned yellow, while others showed red leaves, probably a stress indicator. Sentinel-2 data revealed that spectral indices correlate well with the date of the beginning of senescence, thus suggesting that it would possible to map it. These results already offer evidence that monitoring variability in the dynamics of senescence is possible from satellite remote sensing. Current focus is on the link between canopy colour, as it appears in the webcam imagery, and satellite data.