Comparison of two calibration levels on the simulation of soil water content using nine crop models under different rotation schemes in five European sites.

Diversification of crop rotations is a basic agronomic practice recommended to increase the resilience of agroecosystems, especially in a context of climate change. The majority of crop simulation studies have focused in simulating single crops during singles years. In a long term perspective, it makes more sense to simulate rotations than single crops because they can also characterize the carry-over effects of the previous crop, providing much better arguments for impact and adaptation studies. The aim of this study is to compare two levels of crop model calibration on the projections of soil water availability using nine different crop models (using rotation and single year simulations) in five sites and under different rotation schemes. The low calibration level contained only information related to soil water content and soil mineral N at a date close to sowing day. High calibration contained detailed information about soil parameters, management, so as plant phenology. Data sets from five European sites including ten crops under different crop rotation schemes were used. The targeted variable was soil water content during different periods along the crop season. Results indicate that, for the majority of crops, the high calibration does improve the modelling performance of the dynamics of soil water content for all models when compared with low calibration. When comparing models using rotation or single year simulation, it could be perceived that rotation models can better mimic the long term dynamics of soil water availability. Site conditions also play a role in the quality of the simulation. Overall, models capable of simulating rotation schemes perform better than single year models, if the objective is to assess the soil water dynamic in long term.