This study, developed by UNIMONT , alpine Hub of Milan University, and the Department of Environmental Science and Policy of the University of Milan in the framework of the PNRR Agritech and CIRCAGRIC-GHG projects aims at developing innovative digital agriculture tools and practices at multiple scales for adaptative management of grasslands, with the Val Camonica and Valle del Cervino alpine valleys (Italy) as study.cases and perspective “living laboratory” for improved territorial management of mountain regions.
In the Italian Alps the last decades have seen a significant trend of abandonment and degradation of grazing mountain pastures and meadows, altering valuable ecosystem services, and compromising circularity of agricultural production at farm and regional level. This study, developed by UNIMONT and by the Department of Environmental Science and Policy in the framework of the PNRR Agritech and CIRCAGRIC-GHG projects, aims at developing innovative digital agriculture tools and practices at multiple scales for adaptative management of grasslands, with the Val Camonica and Valle del Cervino alpine valleys (Italy) as study-cases and perspective “living laboratory” for improved territorial management of mountain regions.
At landscape scale, available land-use and land cover (LULC) maps in the Alpine region fail to capture the variability in pasture types and land-covers that reflect abandonment and transition of grasslands to forests, shrubland or degraded conditions. This strongly limits our capacity to monitor and understand grassland degradation processes and their impacts on ecosystem services (e.g., carbon stocking, water cycle regulation, etc.), thus our capacity to support informed decision-making on territorial planning. Therefore, a new LULC map (10m resolution) for the high Val Camonica region (715 km2) was created by applying machine learning classification on Sentinel-1 (radar) and Sentinel-2 (optical) satellite data, combined with a digital elevation model.
Specifically, the thematic map includes 7 pastures typologies with different vegetation composition and productivity, reflecting transitions from more extensive to more intensive use of grasslands. For training and testing, a dataset of 700 photo-interpreted and field-checked ground truth points was used. Preliminary results show a classification accuracy of 82% for the pasture classes. The classification scheme is implemented in an open-source cloud computing environment and can be updated annually. Historical pastoral land-cover loss dynamics have also been evaluated using historical regional land-cover data (i.e., DUSAF 1999 and land-cover map 1980, Lombardy Region), showing a significant decrease in available grazing area and a general decline in pasture quality.
At local scale, we aim at using satellite remote sensing to develop management systems providing timely information on pasture quality and nutritional value during the growing season, to inform adaptative grazing management and promote a more-efficient use of mid to high-elevation alpine pastures.
Specifically, the study is evaluating the potential of last-generation field (ASD Fieldspec4) and satellite (PRISMA, EnMAP, Sentinel 2) remote sensing data to monitor pasture biomass, nutritional status and composition (i.e., green/non-green biomass, proteins, fibers, lignin content). Satellite hyperspectral sensors provide continuous spectral narrow-bands (about 10 nm) in the 400 – 2500 nm range, opening new opportunities for pasture characterization and mapping. Multiple field campaigns have been conducted during summer 2024 on mid-high elevation pastures in Val Camonica and Valle del Cervino coupling hyperspectral reflectance measurements from field and satellite sensors, field agronomic data collection and lab analysis. Predictive models will be developed to (i) understand the potential of different satellite sensors and processing methodologies to provide reliable information on the seasonal dynamics of pasture quantity and quality, (ii) generate intra-seasonal maps that could facilitate decision-making on livestock grazing through the definition of adaptative management units.
For more information: Francesco.fava@unimi.it, Rodolfo.ceriani@unimi.it, francesca.sapio@unimi.it
