• English 
  • Spanish 

The ECOMS User Data Gateway: Towards seasonal forecast data provision and research reproducibility in the era of Climate Services

Journal: Climate Services
Year: 2018   Volume: 9
Initial page: 33   Last page: 43
Status: Published
In this status since: 31 Jan 2018
PDF file: Cofino_2017_CLISER.pdf
Link to PDF: online paper
DOI: 10.1016/j.cliser.2017.07.001

Sectorial applications of seasonal forecasting require data for a reduced number of variables from different datasets, mainly (gridded) observations, reanalysis, and predictions from state-or-the-art seasonal forecast systems (such as NCEP/CFSv2, ECMWF/System4 or UKMO/GloSea5). Whilst this information can be obtained directly from the data providers, the resulting formats, temporal aggregations, and vocabularies may not be homogeneous across datasets. Moreover, different data policies hold for the different datasets, being only some of them publicly available. Therefore, obtaining and harmonizing multi-model seasonal forecast data for sector-specific applications is an error-prone, time consuming task. In order to facilitate this, the ECOMS User Data Gateway (ECOMS-UDG) was developed in the framework of the ECOMS projects as a one-stop shop for climate data. To this aim, the variables required by end users were identified, downloaded from the data providers and locally stored as virtual datasets in a THREDDS Data Server (TDS), implementing fine-grained user management and authorization via the THREDDS Access Portal (TAP). As a result, users can retrieve the subsets best suited to their particular research activities in a user-friendly form using the standard TDS data services. Moreover, an R interface for data access and postprocessing was developed in the form of a bundle of R packages implementing harmonized data access (a single vocabulary), data collocation, bias adjustment and downscaling, and forecast visualization and validation. This provides a unique comprehensive framework, based on a popular scientific computing environments, for end-to-end applications of seasonal predictions , hence favoring reproducibility of results (all packages are publicly available).