Grupo de Meteorología de Santander

Un enfoque multidisciplinar para la meteorología y el clima

Description of the project: pdf
One of the goals of this project is the study of the current perturbations techniques for generating ensembles of climate models. Nowadays, most of the operative weather prediction systems and the seasonal forecasting systems are based on the idea of ensemble methods which account for several integrations from one general circulation model or more. Fluctuations due to errors in the initial conditions (chaos), reduction of variables in the model (noise) and boundary conditions (disorder) plays an important role in the generation of an ensemble. These aspects are analysed here to stablish a proper methodology to characterize the spatiotemporal error growth of ensemble prediction systems. At this point a new perturbation technique based on breeding is introduced in the framework of this project by Primo et al (2008). Logarithmic Bred Vectors (LBV) allows growing vectors with tuneable spatial structure, more (or less) localized. This yields ensembles with different spatiotemporal dynamics (different spread, etc.). The new method increases the diversity of the ensemble and allows the spread to grow faster preserving the model performance in terms of the root mean square error. Consequently, the ensembles can be calibrated for a desired lead time (for instance a shorter forecast range). The method has been first validated using a chain of diffusively coupled Lorenz systems. Moreover it has been found a clear application on ensembles of climate models such as the operative ECMWF weather prediction model.

A second aim is to provide an objective view of the current status of seasonal forecasts at different latitudes. To this end appropriated downscaling methods for ensemble forecasts has been used to perform regional studies using the 40 years of seasonal integrations from the DEMETER project. Seasonal predictability has been analysed in two regions: Perú at tropical latitudes and clearly influenced by El Niño phenomenon and Spain at middle latitudes and far away from the direct influence of ENSO events. It has been found some evidence that accurate local predictions for accumulated seasonal precipitation in Perú can be obtained some months in advance for strong El Niño episodes (Gutierrez et al (2005)). Over Spain the statistical downscaling method improves the results as the skill of the direct forecast increases. The highest skills for precipitation are associated with early and late spring, summer and autumn seasons at 0 and 1 month lead times (Diez et al (2005)).

The study of seasonal predictability over these regions has been extended introducing a new validation method of seasonal forecasts based on intervals (Sordo et al (2008)). This method accounts for the ensemble spread providing an estimation of the uncertainty associated to the forecasts. The interval-based approach was applied to the ECMWF operational seasonal forecast system. High predictability for boreal winter precipitation over Peru during El Niño episodes is found. Predictability over Spain is lower however, the interval-based method is able to uncover some winter precipitation predictability over Spain related to drought episodes. This fact is explained by a known teleconnetion between these negative extreme episodes and La Niña events.

Research at these issues continues in the SEASONAL 2 project also funded by the Spanish government.