Protocol to support model selection and evaluation in a modular crop modelling framework: An application for simulating crop response to nitrogen supply

► A protocol is needed to guide modular framework’s users in the selection of modules. ► The protocol reports on the decisions made during the modelling process. ► The use of uncertainty matrix prevents from misusing a modular crop modelling framework. ► Model building should be seen as an iterative process. Crop models require different structures for different applications. Modular and flexible crop modelling frameworks, such as the recently developed agricultural production and externalities simulator (APES), support the change of model structure. However, the assembly of different modules to create a model may not always result in the best model structure. We developed and tested a protocol for a systematic selection and evaluation of a crop growth model structure. The novelty of the presented protocol relies on a throughout analysis of the different modelling approaches (modules) and on how to assemble them to create new modelling solutions (i.e. model). We use a case study to demonstrate that we can explicitly express and test the different assumptions behind the choice of a specific modelling approach. Our case study refers to the simulation of crop growth in response to nitrogen management and the importance of an accurate simulation of the nitrogen uptake. Applying the proposed protocol, we identify the need to improve the initially selected nitrogen mineralisation module. We conclude that not only is the protocol suitable to provide guidance for systematic testing of different crop processes modelled, but also its use highlights the importance of the documentation of the modelling process and of the clarification of the uncertainty associated with the model structure.