Innovations and Lessons Learned Developing the USDA Long-Term Agroecosystem Research Network Common Observatory Data Repository

Abstract:

The objective for the USDA Long-Term Agroecosystem Research (LTAR) network Common Observatory Repository (CORe) is to provide data management services including archive, discovery, and access for consistently observed data across all 18 nodes. LTAR members have an average of 56 years of diverse historic data. Each LTAR has designated a representative ‘permanent’ site as the location’s common meteorological observatory. CORe implementation is phased, starting with meteorology, then adding hydrology, eddy flux, soil, and biology data. A design goal was to adopt existing best practices while minimizing the additional data management duties for the researchers. LTAR is providing support for data management specialists at the locations and the National Agricultural Library is providing central data management services.

Maintaining continuity with historical observations is essential, so observations from both the legacy and new common methods are included in CORe. International standards are used to store robust descriptive metadata (ISO 19115) for the observation station and surrounding locale (WMO), sensors (Sensor ML), and activity (e.g., re-calibration, locale changes) to provide sufficient detail for novel data re-use for the next 50 years. To facilitate data submission a simple text format was designed. Datasets in CORe will receive DOIs to encourage citations giving fair credit for data providers. Data and metadata access are designed to support multiple formats and naming conventions. An automated QC process is being developed to enhance comparability among LTAR locations and to generate QC process metadata. Data provenance is maintained with a permanent record of changes including those by local scientists reviewing the automated QC results. Lessons learned so far include increase in site acceptance of CORe with the decision to store data from both legacy and new common methods. A larger than anticipated variety of currently used methods with potentially significant differences for future data use was found. Cooperative peer support among locations with the same sensors coupled with central support has reduced redundancy in procedural and data documentation.