Caracterización del viento en la Península Ibéricaobservaciones y simulaciones regionales

Abstract

The aim of this Thesis is (1) to characterize the surface wind over the Iberian Peninsula through the use of observations and Regional Climate Models (RCMs), (2) to analyze the simulation errors and (3) to propose and implement improvements in the model physics for enhancing the wind reproduction. The methodology consists of various procedures. First, a quality-controlled database of surface wind (speed and direction) has been built. This database has no precedent and is composed of 514 hourly series covering the Iberian Peninsula for the period 2002-2007. Second, RCMs are used to assess the wind climatology and to overcome the limitations of the observations. A regional climate simulation covering the Iberian Peninsula with 10 km of spatial resolution during 1959-2007 is performed. Third, multivariate techniques (clustering analysis and principal components analysis) are applied to facilitate the wind characterization and validate the simulation at regional scale. The Iberian Peninsula is split in several regions attending to similar wind speed variability. Fourth, two additional simulations are run with WRF model during the observational period (2002-2007) in order to study the parameterization effect of the subgrid topographic friction. Finally, the term of the subgrid topographic fiction is modified in WRF in order to consider the influence of the atmospheric stability in the representation of surface wind. The results indicate that there is strong spatial wind variability over the Iberian Peninsula. This is characterized by the annual cycle, wind rose, wind speed histograms, as well as the temporal correlations of the mean regional series. In addition, the wind climate heterogeneity can be described as the interaction of the orography with the main atmospheric circulation types affecting the Iberian Peninsula. MM5 simulation is able to reproduce the main wind features of the wind over the Iberian Penisula, providing an important added value with respect to the driving conditions. The simulation has been used for several studies and applications. The limitations of the observational database have been inspected. The observational period is not long enough to describe the wind climatology in the western Iberian Peninsula and the inclusion more weather stations in some mountainous regions would be desirable. The simulation and observations are used to evaluate the wind power resource. A public web mapping tool for facilitating the access to wind information has been developed. In addition, the long period simulated has allowed the analysis of the wind variability during the last five decades. A decline of the wind speed during winter has occurred, associated to a decrease of cyclonic situations. On the other hand, the wind speed has increased during summer due to an enhancement of thermal circulations. The major caveat with the MM5 and WRF simulations was a systematic wind speed overestimation. The inclusion of the parameterization of the subgrid topographic effects in WRF reduces the wind speed bias. However, this scheme underestimates the wind speed during windy situations and the amplitude of the diurnal cycle, being the main disagreements during the day. The relationship of the wind speed bias with the convective velocity, the turbulent kinetic energy and the height of the boundary layer points out that the induced subgrid topographic friction is influenced by the atmospheric stability conditions. The modulation of the frictional term as a function of the stability parameters improves the reproduction of the wind speed daily cycle. However the improvement depends on the time of the year. This Thesis contributes to deepen in the knowledge of the variability of surface winds over the Iberian Peninsula by means of observations and regional climate models. This work also represents and advance in atmospheric simulations because drawbacks associated with the simulated winds are improved by implementing new physics schemes in mesoscale models.