From instantaneous to continuous: Using imaging spectroscopy and in situ data to map two productivity-related ecosystem services

Spatially well-informed decisions are essential to sustain and regulate processes and ecosystem services (ES), and to maintain the capacity of ecosystems to supply services. However, spatially explicit ES information is often lacking in decision-making, or exists only as ES maps based on categorical land cover data. Remote sensing (RS) opens new pathways to map ES, in particular biophysical ES supply. We developed an observation-based concept for spatially explicit and continuous ES mapping at landscape scale following the biophysical part of the ES cascade. We used Earth observations in combination with in situ data to map ecosystem properties, functions, and biophysical ES supply. We applied this concept in a case study to map two ES: carbon dioxide regulation and food supply. Based on Earth observations and in situ data, we determined the ecosystem property Sun-Induced chlorophyll Fluorescence (SIF) to indicate ecosystem state and applied scaling models to estimate gross primary production (GPP) as indicator for ecosystem functioning and consequently carbon dioxide regulation and food supply as ES. Resulting ES maps showed heterogeneous patterns in ES supply within and among ecosystems, which were particularly evident within forests and grasslands. All investigated land cover classes were sources of CO2, with averages ranging from -66 to -748 g C m(-2) yr(-1), after considering the harvest of total above ground biomass of crops and the storage organ, except for forest being a sink of CO2 with an average of 105 g C m(-2) yr(-1). Estimated annual GPP was related to food supply with a maize grain yield average of 9.5 t ha(-1) yr(-1) and a sugar beet root yield of 110 t ha(-1) yr(-1). Validation with in situ measurements from flux towers and literature values revealed a good performance of our approach for food supply (relative RMSE of less than 23%), but also some over- and underestimations for carbon dioxide regulation. Our approach demonstrated how RS can contribute to spatially explicit and continuous ES cascade mapping and suggest that this information could be useful for environmental assessments and decision-making in spatial planning and conservation.