An overview of AmeriFlux data products and methods for data acquisition, processing, and publication

Friday, 19 December 2014
Gilberto Pastorello, Lawrence Berkeley National Lab, Emeryville, CA, United States, Cristina Poindexter, University of California Berkeley, Berkeley, CA, United States, Deb Agarwal, LBNL, Berkeley, CA, United States, Dario Papale, Tuscia University, Department for Innovation in Biological, Agro-food and Forest systems (DIBAF), Viterbo, Italy, Catharine van Ingen, Lawrence Berkeley National Laboratory, Berkeley, CA, United States and Margaret S Torn, Berkeley Lab/UC Berkeley, Berkeley, CA, United States
The AmeriFlux network encompasses independently managed field sites measuring ecosystem carbon, water, and energy fluxes across the Americas. In close coordination with ICOS in Europe, a new set of fluxes data and metadata products is being produced and released at the FLUXNET level, including all AmeriFlux sites. This will enable continued releases of global standardized set of flux data products. In this release, new formats, structures, and ancillary information are being proposed and adopted. This presentation discusses these aspects, detailing current and future solutions. One of the major revisions was to the BADM (Biological, Ancillary, and Disturbance Metadata) protocols. The updates include structure and variable changes to address new developments in data collection related to flux towers and facilitate two-way data sharing. In particular, a new organization of templates is now in place, including changes in templates for biomass, disturbances, instrumentation, soils, and others. New variables and an extensive addition to the vocabularies used to describe BADM templates allow for a more flexible and comprehensible coverage of field sites and the data collection methods and results. Another extensive revision is in the data formats, levels, and versions for fluxes and micrometeorological data. A new selection and revision of data variables and an integrated new definition for data processing levels allow for a more intuitive and flexible notation for the variety of data products. For instance, all variables now include positional information that is tied to BADM instrumentation descriptions. This allows for a better characterization of spatial representativeness of data points, e.g., individual sensors or the tower footprint. Additionally, a new definition for data levels better characterizes the types of processing and transformations applied to the data across different dimensions (e.g., spatial representativeness of a data point, data quality checks applied, and differentiation between measured data and data from models that use process knowledge). We also present an expanded approach to versions of data and data processing software, with stable and immutable data releases, but also pre-release versions to allow evaluation and feedback prior to a stable release.