Multispectral and SAR satellite data to assess drought impact on the productivity of mountain grasslands in the European Alps

Drought events occur more and more often in the Alps, endangering the welfare of mountain agriculture. In this context, risk management instruments, like insurance, can help agricultural systems cope with production shortcomings and thus increase their economic resilience, prevent land abandonment, and maintain their functioning over time. In this work, we focus on Trentino-South Tyrol, in north-eastern Italy, where mountain grasslands play an important economic role as they provide forage for livestock farming and a place of recreation for tourists. In addition, they contribute to many ecosystem services including climate regulation, biodiversity and landscape conservation, and soil protection. In collaboration with stakeholders from the agricultural sector, we developed an index of productivity, the Grassland Production Index (GPI), which can be used at the end of the growing season to assess yield losses due to drought events. GPI is estimated from meteorological data and leaf area index (LAI) derived from Sentinel-2 multispectral data. LAI and GPI are validated by field measurements of LAI and dry matter yield covering two growing seasons at eight test sites per year. This is achieved by a well-established and replicable data collection protocol. The validation of the Sentinel-2 LAI with ground measurements showed an RMSE of 0.92 [m2 m−2] and an R2 of 0.81 over all the measurement sites. A comparison between GPI and yield showed, on average, an R2 of 0.56 at the pixel scale and an R2 of 0.74 at the parcel scale. Based on these promising validation results, the index was applied to estimate the insurance payments for four farms. An advanced version of GPI is under development in which we improve the Sentinel-2 LAI time series by a data fusion approach. Here, missing LAI values due to cloud coverage are estimated by machine learning algorithms with input features calculated from backscattering and soil moisture derived from Sentinel-1 SAR data. The application of the enhanced GPI for insurance purposes at the regional scale is foreseen at the end of the 2024 growing season. This work presents a real case study using GPI for drought impact assessment and investigates the potential of fusing optical and SAR data to improve the estimation of GPI.