Large-scale mechanisms influencing the winter Romanian climate variability

The mechanisms by which the large-scale circulation influences the Romanian climate variability for the period 1946-1995 are discussed, The seasonal mean temperature and seasonal precipitation at 14 Romanian stations represent the regional climate. The large-scale circulation is represented by the sea level pressure (SLP) and geopotential height (H500) over the Atlantic-European region. The link between regional and large-scale parameters is studied using the Canonical Correlation Analysis (CCA). CCA is applied separately to pairs of one large-scale parameter and one local parameter, as well as to the combined vector of large-scale and local parameters. Changes in the local parameters are studied in connection to changes in the large-scale circulation. The statistical downscaling model based on the most important CCA time series is built. This analysis was done for the winter season. It was concluded that the link with H500 and SLP is strong for both local parameters. For temperature, the skill of the statistical downscaling model is significantly higher by using the combined vector of large-scale predictors and the combined vector with precipitation as the predictand. In the case of precipitation the skill of the model is not significantly improved when the combined vectors are used. This result lead to the conclusion that precipitation variability is governed by a common large-scale circulation mechanism. The viability of this conclusion is also proved by the changes in the H500 and SLP patterns around 1970, which are in agreement with changes in the precipitation field, The temperature variability is influenced by more signals, the local influence of precipitation and global warming also being very important for this season.