This study deals with an analysis of the performance of a general circulation model (GCM) (HadCM2) in reproducing the large-scale circulation mechanisms controlling Swedish precipitation variability, and in estimating regional climate changes owing to increased CO 2 concentration by using canonical correlation analysis (CCA). Seasonal precipitation amounts at 33 stations in Sweden over the period are used. The large-scale circulation is represented by sea level pressure (SLP) over the Atlantic-European region.The link between seasonal Swedish precipitation and large-scale SLP variability is strong in all seasons, but especially in winter and autumn. For these two seasons, the link is a consequence of the North Atlantic Oscillation (NAO) pattern. In winter, another important mechanism is related to a cyclonic/anticyclonic structure centred over southern Scandinavia. In the past century, this connection has remained almost unchanged in time for all seasons except spring. The downscaling model that is built on the basis of this link is skilful in all seasons, but especially so in winter and autumn. This observed link is only partially reproduced by the HadCM2 model, while large-scale SLP variability is fairly well reproduced in all seasons. A concept about optimum statistical downscaling models for climate change purposes is proposed. The idea is related to the capability of the statistical downscaling model to reproduce low frequency variability, rather than having the highest skill in terms of explained variance. By using these downscaling models, it was found that grid point and downscaled climate signals are similar (increasing precipitation) in summer and autumn, while in winter, the amplitudes of the two signals are different. In spring, both signals show a slight increase in the northern and southern parts of Sweden.

Empirical downscaling procedures relate large-scale atmospheric features with local features such as station rainfall in order to facilitate local scenarios of climate change. The purpose of the present paper is twofold: first, a downscaling technique is used as a diagnostic tool to verify the performance of climate models on the regional scale; second, a technique is proposed for verifying the validity of empirical downscaling procedures in climate change applications. The case considered is regional seasonal precipitation in Romania. The downscaling model is a regression based on canonical correlation analysis between observed station precipitation and European-scale sea level pressure (SLP). The climate models considered here are the T21 and T42 versions of the Hamburg ECHAM3 atmospheric GCM run in “time-slice” mode. The climate change scenario refers to the expected time of doubled carbon dioxide concentrations around the year 2050. The downscaling model is skillful for all seasons except spring. The general features of the large-scale SLP variability are reproduced fairly well by both GCMs in all seasons. The climate models reproduce the empirically determined precipitation–SLP link in winter, whereas the observed link is only partially captured for the other seasons. Thus, these models may be considered skillful with respect to regional precipitation during winter, and partially during the other seasons. Generally, applications of statistical downscaling to climate change scenarios have been based on the assumption that the empirical link between the large-scale and regional parameters remains valid under a changed climate. In this study, a rationale is proposed for this assumption by showing the consistency of the 2 × CO2 GCM scenarios in winter, derived directly from the gridpoint data, with the regional scenarios obtained through empirical downscaling. Since the skill of the GCMs in regional terms is already established, it is concluded that the downscaling technique is adequate for describing climatically changing regional and local conditions, at least for precipitation in Romania during winter.

The variability of winter mean precipitation as observed at 14 Romanian rain gauge stations from 1901–1988 is examined. Pettitt's statistic is used to detect changes of regimes in the time series. Almost all stations exhibit a systematic decrease (“downward shift”) at about 1969. Furthermore, upward shifts are identified for the southwestern stations at about 1933, and a downward shift in the mid 1920's in the northwest. An upward shift at about 1919 for the Bucharest station is likely determined by the urbanisation effect. These systematic changes are shown to be real and not an artifact due to inhomogeneities in the precipitation data in a two‐step procedure. First, the precipitation field and the European‐scale sea‐level air‐pressure field are related to each other through a Canonical Correlation Analysis (CCA). Two relevant pairs of characteristic patterns are found. In a second step, the CCA‐coefficients of these two pairs are studied with Pettitt's statistic. In both pairs of time series, simultaneous change points are found in the precipitation and in the pressure‐related coefficients. The 1933 and 1969 change points are related to a change of the southwesterly flow which brings moist Mediterranean air to Romania. The mid‐1920s change point is triggered by changes in the frequency or intensity of the north‐westerly circulation. As a byproduct, we found that Pettitt's statistic is sensitive to the presence of trends and serial correlation so that its use for statistical hypothesis testing is limited. Therefore, we have used Pettitt's statistic only as an explanatory tool.

The main characteristics of spatial and temporal variability of the precipitation regime in Sweden were studied by using the long-term monthly precipitation amount (1890-1990) at 33 stations. The data were filtered by using Empirical Orthogonal Function (EOF) analysis, which provides principal modes of both spatial variability and time coefficient series describing the dominant temporal variability. Canonical correlation analysis (CCA) was used to reveal association between the atmospheric circulation and the characteristics of the climate variability.