Forest traits from PRISMA spaceborne imagery

Forest ecosystems, spanning approximately one-third of the Earth's landmass, play a crucial role in providing essential ecosystem services. However, their extension and condition are under threat due to the impacts of climate change. While remote sensing holds the potential to assess the condition and functionality of global forests, challenges in methodology and technology hinder the accurate quantitative estimation of forest traits from spaceborne observations. The emergence of new-generation satellites and advanced retrieval techniques offers the prospect of overcoming these obstacles, yet the potential of both the data and models requires further evaluation. In this contribution, we focused on retrieving forest traits from PRISMA hyperspectral spaceborne imagery employing machine learning regression models as well as hybrid approaches. The area we selected for this study is the Ticino Park, a mid-latitude forest located in northern Italy along the Po river. We conducted an intensive field campaign in the park in the summer of 2022 in conjunction with four PRISMA overpasses to collect trait samples for the calibration and validation of the retrieval schemes. The results obtained highlighted the capability of PRISMA images and retrieval models to precisely quantify Leaf Area Index (LAI) (R²=0.91, nRMSE=8.3%), Leaf Water Content (LWC) (R²=0.97, nRMSE=4.7%) and Leaf Mass per Area (LMA) (R²=0.95, nRMSE=5.6%) in forest ecosystems. Less performing but still promising results were obtained for Leaf Chlorophyll Content (LCC) (R²=0.44, nRMSE=18.3%) and Leaf Nitrogen Content (LNC) (R²=0.63, nRMSE=14.2%). The comparison of the trait values in June and early September revealed a significant decline in both leaf biochemistry and LAI, which can be traced back to the stress induced in the Ticino Park by the severe drought that hit Europe in the summer of 2022. This underscores the valuable role of hyperspectral spaceborne imagery and new generation models for monitoring forest conditions.