European-wide simulations of croplands using an improved terrestrial biosphere model: Phenology and productivity

Aiming at producing improved estimates of carbon source/sink spatial and interannual patterns across Europe (35% croplands), this work combines the terrestrial biosphere model Organizing Carbon and Hydrology in Dynamic Ecosystems (ORCHIDEE), for vegetation productivity, water balance, and soil carbon dynamics, and the generic crop model Simulateur Multidisciplinaire pour les Cultures Standard (STICS), for phenology, irrigation, nitrogen balance, and harvest. The ORCHIDEE-STICS model, relying on three plant functional types for the representation of temperate agriculture, is evaluated over the last few decades at various spatial and temporal resolutions. The simulated leaf area index seasonal cycle is largely improved relative to the original ORCHIDEE simulating grasslands, and compares favorably with remote-sensing observations (correlation doubles over Europe). Crop yield is derived from annual net primary productivity and compared with wheat and grain maize harvest data for five European countries. Discrepancies between 30 year mean simulated and reported yields are large in Mediterranean countries. Interannual variability amplitude expressed relative to the mean is reduced toward the observed variability (approximate to 10%) when using ORCHIDEE-STICS. Overall, this study highlights the importance of accounting for the specific phenologies of crops sown both in winter and in spring and for irrigation applied to spring crops in regional/global models of the terrestrial carbon cycle. Limitations suggest to account for temporal and spatial variability in agricultural practices for further simulation improvement.