The model-as-a-resource paradigm for geoscience digital ecosystems

A long-term goal of environmental science and Earth observation is to enable the creation of a “Model Web” of semantically interconnected data and models. Geospatial models are usually exposed on the Web as services accessible through heterogeneous interfaces. However, such services, which represent instances of the paradigm called Model-as-a-Service (MaaS), cannot be easily exploited beyond their original use as defined by the service provider. To overcome this important limitation and better support transparency, reproducibility, replicability and reusability of the model (following the Open Science paradigm), we investigated the adoption of a Model-as-a-Resource (MaaR) approach, in which a model is considered a generic digital resource that, as such, can play different roles in different potential use cases. The proposed MaaR framework can play an important enabling role in the realization of those digital ecosystems that generate environmental knowledge. The main challenges and opportunities are discussed in the manuscript.