USING UAVs AND VENμS TO CHARACTERIZE THE PHENOLOGY OF MEDITERRANEAN WOODY SPECIES ACROSS SPATIAL SCALES

Due to spectral similarity, discrimination between woody Mediterranean species via remote sensing is not straightforward. However, considering plants phenology potentially allows identification at the individual plant level. The VENμS satellite has a multispectral sensor, with high spatial (5.3m) and temporal resolutions (two days revisit time), increasing its potential for successful acquisitions of cloud-free data and enabling a more accurate detection of phenological events. This study aims to examine the compatibility between phenology patterns of woody Mediterranean species obtained from different platforms, over various spatial resolutions. We therefore explore the ability of weekly consumer-grade camera (VIS-NIR), monthly multispectral UAV (0.13m), and VENμS satellite images taken throughout 2018 from a research site located in the Judaean mountains (Mata), to monitor phenology. After illumination corrections, six vegetation indices (VI) values were calculated from the ground camera data (57 dates), followed by matching of a locally weighted scatterplot smoothing function (LOESS) and outlier removal, resulting in 691 individual plants of 12 common species. Due to low percentage of outliers the excess green (ExG) VI time series was selected to represent the phenology from ground observations. The ExG had the highest correlation compared to the other VIs computed, between ground-based measurements and UAV data (11 out of 12 species, R>0.75; p-value<0.01). The ability to detect phenology using sensors with coarser spatial resolution will be explored by resampling the UAV data (cell resizing to 0.3m and 1.25m) and computing VIs out of VENμS imagery, followed by evaluating the different scales correlation to the near-ground derived phenology.