Santander Meteorology Group

A multidisciplinary approach for weather & climate

Description of the project: pdf
This project represents an extension to the advances in seasonal forecasting addressed in a previous project (SEASONAL, CGL-2004-02652/CLI). SEASONAL 2 is focused on the analysis and characterization of the spatial-temporal dynamics of multimodel ensemble predictions and on the validation of multimodel ensemble seasonal forecasts from the current European project ENSEMBLES. The methodology developed in SEASONAL to characterize and distinguish different types of forecasts and numerical models according to a spatial-temporal analysis is here applied to the GCM outputs from the European project ENSEMBLES and also to those from its predecessor DEMETER. A novel approach to characterize and
graphically represent the spatiotemporal evolution of ensembles is introduced using a simple diagram. The MVL diagram (Mean-Variance of Logarithms) intuitively represents the interplay and evolution of the first two moments of the logarithmic transformed values: the mean which is associated with the exponential growth in time and the variance which accounts for the spatial correlation and localization of fluctuations. The MVL diagram has been applied to toy models and numerical weather prediction systems (Gutierrez et al 2008) being observed that it uncovers useful information about the spatiotemporal dynamics of the ensemble. The MVL diagram has been also applied to multi-model ensembles from the DEMETER project analyzing the effect of both the initialization procedures and the model formulation differences (Fernández et al 2008). It is shown that the shared building blocks (atmospheric and ocean components) impose similar dynamics among different models and, thus, contribute to poorly sampling the model formulation uncertainty. This dynamical similarity should be taken into account, at least as a pre-screening process, before applying any objective weighting method.

Previous results from SEASONAL shown some predictability in Spain therefore, this project is focused on extending the seasonal forecast verification studies to Europe. The number of variables considered is extended to precipitation, temperature and wind and more statistical downscaling methods are applied. The influence of ENSO events as
a potential source of conditional predictability is first assessed. A particular study has been already developed over Spain using a high resolution gridded data set. Skillful seasonal predictions provided by the DEMETER project are found in partial agreement with the observed teleconnections derived from historical records (Frias et al 2008). Skillful dry events over the South and the Mediterranean coast and warm events in the South-East during spring are found related to El Niño phenomena. On the other hand, La Niña events drive higher predictability in winter associated to dry events over the Western part and to warm events in summer over the South and the Mediterranean coast of Spain.

Furthermore, sensitivity studies are also carried on taking into account the number of ensemble members, the reanalysis data used or the observations considered (station data or grid data).