Data Repositories for Meta-Analyses and Support of Models.

Currently available repositories intended to manage and curate data exhibit a wide range in characteristics that control their usefulness for modeling and meta-analyses. These characteristics relate to: completeness, accessibility, spatial/temporal coverage, quality control, standardization of measuring and reporting protocols, documentation, etc. For example, the US Trace Gas Network contains soil GHG flux data but it is currently not publically available and contains little ancillary data needed for model inputs and complete model testing. In contrast, the GRACEnet/REAP data management system houses complete agro-ecosystem model driver and testing data that are publically available. The Ecosystem Model-Data Intercomparison (EMDI) project compiled Net Primary Productivity (NPP) measurements and model driver data for numerous sites around the world. The data were partitioned onto classes based on degree of documentation and completeness of measurements and were used to compare C cycle models. The Long-Term Intersite Decomposition Experiment Team (LIDET) collected and compiled litter bag data for most of the world's biomes. Data have been used extensively for meta-analyses to determine the chemical and environmental controls on decomposition rates. Recently, international groups such as the Global Research Alliance on Agricultural Greenhouse Gases (GRA) and the Agricultural Model Intercomparison and Improvement Project (AgMIP) have compiled model driver and testing data sets, as well as detailed procedures to conduct simulations to ensure that model comparisons are objective. The pros and cons of the data systems and model testing procedures developed for the projects above will be further discussed.