Process-based greenhouse climate models : Genealogy, current status, and future directions

CONTEXT: Process-based greenhouse climate models are valuable tools for the analysis and design of greenhouse systems. A growing number of greenhouse models are published in recent years, making it difficult to identify which components are shared across models, which are new developments, and what are the objectives, strengths and weaknesses of each model. OBJECTIVE: We present an overview of the current state of greenhouse modelling by analyzing studies published between 2018 and 2020. This analysis helps identify the key processes considered in process-based greenhouse models, and the common approaches used to model them. Moreover, we outline how greenhouse models differ in terms of their objectives, complexity, accuracy, and transparency. METHODS: We describe a general structure of process-based greenhouse climate models, including a range of common approaches for describing the various model components. We analyze recently published models with respect to this structure, as well as their intended purposes, greenhouse systems they represent, equipment included, and crops considered. We present a model inheritance chart, outlining the origins of contemporary models, and showing which were built on previous works. We compare model validation studies and show the various types of datasets and metrics used for validation. RESULTS AND CONCLUSIONS: The analysis highlights the range of objectives and approaches prevalent in greenhouse modelling, and shows that despite the large variation in model design and complexity, considerable overlap exists. Some possible reasons for the abundance of models include a lack of transparency and code availability; a belief that model development is in itself a valuable research goal; a preference for simple models in control-oriented studies; and a difference in the time scales considered. Approaches to model validation vary considerably, making it difficult to compare models or assess if they serve their intended purposes. We suggest that increased transparency and availability of source code will promote model reuse and extension, and that shared datasets and evaluation benchmarks will facilitate model evaluation and comparison. SIGNIFICANCE: This study highlights several issues that should be considered in greenhouse model selection and development. Developers of new models can use the decomposition provided in order to present their models and facilitate extension and reuse. Developers are encouraged to reflect on and explicitly state their model's range of suitability, complexity, validity, and transparency. Lastly, we highlight several steps that could be taken by the greenhouse modelling community in order to advance the field as a whole.