Comparing radiative transfer model-based LAI retrieval with in-situ observations and mechanistic modelling for grassland growth assessment

Abstract

Grasslands cover a significant portion of Switzerland’s landscape, primarily serving for livestock production, but also providing many ecosystem services like safeguarding biodiversity, habitat provision, carbon storage or water purification. Yet, through exposure to climate change, but also intensive land use such as frequent mowing and intensive grazing grasslands are increasingly threatened. To evaluate the state of grasslands and optimize sustainable management practices, it is necessary to understand their ecological state, the management strategies and use intensity they're exposed to. A reliable data basis is a prerequisite for an accurate assessment. However, the acquisition of ground-field data is a costly and time-consuming process, and often requires financial resources and human capacity.   Satellite data may provide a cost-effective alternative. The derivation of physical based quantities like the Leaf Area Index (LAI) through Radiative Transfer Models (RTM) has shown great potential to estimate several biophysical and biochemical plant traits from spectral data. This contribution presents a method to estimate grass growth using satellite data time series along with an RTM inversion-based LAI retrieval approach. The results are compared to in-situ observations and results from a mechanistic model. The methodology includes 1) suitable parameterization of the RTM for grasslands and generation of a spectral library, 2) training of the retrieval algorithm (neural network/random forest), and 3) the extraction of LAI time series from satellite images to compute LAI progress over time. We evaluate this method by comparing the results to a grass growth curve computed by the mechanistic model ModVege as well as in-situ data from multiple sites in Switzerland.   The investigation of LAI retrieved from satellite data and RTM inversion for grassland growth assessment provides valuable insights into optimizing grassland management practices. These findings are further utilized to improve the long-term sustainability of grasslands in the face of changing environmental conditions.