|Title||The observation feedback archive for the ICOADS and ISPD data sets|
|Series/Collection||ERA Report Series|
|Authors||Hersbach, H, Poli, P, Dee, DP|
|Event Series/Collection||ERA Report|
|Place of publication||Shinfield Park, Reading|
This document describes the import of the ISPDv2.2, ISPDv3.2.6, and the ICOADSv2.5 historical datasets from their original formats into the observation feedback archive (OFA) at the European Centre for Medium-RangeWeather Forecasts (ECMWF) which includes information on their usage into the NOAA/CIRES 20th century (20CR) and ECMWF ERA-20C reanalyses. Both ISPD data sets contain observations for surface pressure and mean sea level pressure, world-wide, from 1768 to 2010. ICOADSv2.5, which ranges back from 1662, contains marine observations for a number of geophysical quantities. Although it is being extended in near real time, the OFA archive at ECMWF covers until December 2011. The OFA is the unified data model at ECMWF for observations that have been passed through the data-assimilation system. Key element is the addition of feedback, embracing collocated model values and information on data usage. Three feedback archives are presented. The first OFA regards ISPDv2.2 with feedback from 20CR that had used this data set, and which is available in the data set itself. The second OFA regards ICOADSv2.5, which does not contain feedback, but at ECMWF model information from 20CR was added by an offline collocation with available model fields, instead. The third data set embraces feedback from the ERA-20C analysis for the data sets it was based on, i.e., the later version ISPDv3.2.6 and ICOADSv2.5. Although the ISPDv2.2, ICOADSv2.5 and ERA-20C OFA's contain many overlaps, their attached feedback differs. The format of OFA is the simplified representation of the ECMWF observation data base (ODB), which was introduced at ECMWF recently to enable archiving into the ECMWF Meteorological Archival and Retrieval System (MARS). The datasets presented here are the first historical datasets that were ingested into this format and made publicly available on the ECMWF data server. A detailed description is provided on which elements in the native datasets are mapped where into ODB.