Verification of the ECMWF ensemble forecasts of wind speed against analyses and observations

Pierre Pinson, Renate Hagedorn

    Research output: Contribution to journalJournal articleResearchpeer-review

    680 Downloads (Pure)

    Abstract

    A framework for the verification of ensemble forecasts of near-surface wind speed is described. It is based on existing scores and diagnostic tools, though considering observations from synoptic stations as reference instead of the analysis. This approach is motivated by the idea of having a user-oriented view of verification, for instance with the wind power applications in mind. The verification framework is specifically applied to the case of ECMWF ensemble forecasts and over Europe. Dynamic climatologies are derived at the various stations, serving as a benchmark. The impact of observational uncertainty on scores and diagnostic tools is also considered. The interest of this framework is demonstrated from its application to the routine evaluation of ensemble forecasts and to the assessment of the quality improvements brought in by the recent change in horizontal resolution of the ECMWF ensemble prediction system. The most important conclusions cover (1) the generally high skill of these ensemble forecasts of near-surface wind speed when evaluated at synoptic stations, (2) the noteworthy improvement of scores brought by the change of horizontal resolution, and, (3) the scope for further improvements of reliability and skill of wind speed ensemble forecasts by appropriate post-processing.
    Original languageEnglish
    JournalMeteorological Applications
    Volume19
    Issue number4
    Pages (from-to)484–500
    ISSN1350-4827
    DOIs
    Publication statusPublished - 2012

    Keywords

    • Observational uncertainty
    • Benchmark
    • Scores
    • Skill
    • Calibration
    • Probabilistic forecasting
    • Energy application

    Fingerprint

    Dive into the research topics of 'Verification of the ECMWF ensemble forecasts of wind speed against analyses and observations'. Together they form a unique fingerprint.

    Cite this