Archiving Terrestrial Biogeochemistry Models

Archiving environmental data products has become recognized as a vital research practice: it improves our ability to reproduce results and perform additional analyses, while saving the cost of redundant data collection activities. The same rationale applies to archiving numerical models. Archived models will provide the methodological detail of numerical modeling studies to recreate published modeling results, enabling the synthesis of results across modeling studies and the investigation of new hypotheses. In addition, archived models will allow determination of uncertainties for comparison with results from other models in assessment / policy studies. The model source code will also allow others to see how models treat individual processes. We are creating a two-tiered archive for numerical models. The first tier supports the storage and retrieval of benchmark model versions, and the second tier supports the association of published research results with specific model implementations. For both tiers, we present a set of recommended "best practices" aimed at raising the standards for reproducibility in numerical modeling studies through the use of a dedicated archive for numerical models and modeling studies. The model archive contains comprehensive model documentation, input files, code version, output files, and output analysis approaches or software used to produce tables and figures for a particular publication. The expectation is that the model archive will be a resource for experienced modelers.