An extensible, multi-model software library for simulating crop growth and development

The availability of different crop models and of a variety of techniques to evaluate their behaviour led to a change of paradigm in crop models use. Modellers are now looking beyond the idea of groups of users and developers grounded on a specific model, and international initiatives focusing on model improvement basing on intercomparison and knowledge sharing are recently catalysing the attention of the international community. Also, the analysis under environmental conditions of no adaptation by crops, such as the ones of climate change scenarios, demand for extension of crop models to account for extreme events, diseases and pests impact, difficult to implement into legacy code. The Crop Models Library (CropML) is a framework-independent MS .NET software component where different pure (e.g., WOFOST, CropSyst, WARM), hybrid and new modelling solutions for crop growth and development are implemented following a fine level of granularity, according to a high level software architecture. CropML can be extended by third parties and is distributed at no cost with a software development kit, including documentation of code and algorithms and sample applications. CropML provides modellers with an environment favouring the hybridization of models with parts from others, the evolution of existing approaches, and the possibility of analysing and easily comparing diverse modelling solutions. As an example, a new generation of SUCROS-type models has been developed and included in the component. Comparison of the standard and of the new version of the WOFOST model carried out using data from rice field experiments revealed an increase in accuracy and robustness with less than half of the parameters used by the standard version of the model. These results support the idea that high-level technology for models formalization can favour the development of the models themselves.