Open AccessInterval‐based statistical validation of operational seasonal forecasts in Spain conditioned to El Niño–Southern Oscillation events
Author(s) -
Sordo C.,
Frías M. D.,
Herrera S.,
Cofiño A. S.,
Gutiérrez J. M.
Publication year - 2008
Publication title -
journal of geophysical research: atmospheres
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.67
H-Index - 298
eISSN - 2156-2202
pISSN - 0148-0227
DOI - 10.1029/2007jd009536
Subject(s) - predictability , teleconnection , climatology , forecast skill , environmental science , middle latitudes , el niño southern oscillation , meteorology , downscaling , standard deviation , statistics , geography , mathematics , precipitation , geology
As opposed to the tropics, operational seasonal forecasting systems have shown little or no skill in European midlatitudes. In this paper we explore the potential source of predictability in this region given by El Niño–Southern Oscillation (ENSO) events; in particular we analyze winter rainfall in Spain. First, we apply a simple statistical method to assess the teleconnections between rainfall records in 123 gauges over Spain and ENSO events during the last 40 years. A significant teleconnection for dry winter episodes is found associated with La Niña events, extending the results obtained in previous studies. Then, we adapt the statistical method to perform operational seasonal forecasts validation conditioned to ENSO events; in particular we consider a state‐of‐the‐art operational model, the System2 from ECMWF. The validation method defines a forecast interval to account for the ensemble spread, and applies a simple skill measure based on the proportion of hits (observations falling into the forecast interval) compared with a random forecast. As a result, we uncover the significant skill of operational seasonal predictions for reproducing the dry winter episodes associated with La Niña events (a window of opportunity for operational seasonal forecast in midlatitudes). Finally, the results are improved using statistical downscaling methods and some sensitivity studies are conducted. The analysis presented in this paper can be extended to other regions under the influence of any seasonal predictability‐driving factor.