An attempt is made to characterize the mesoscale climatological structures of the daily maximum and minimum temperature anomalies (with respect to monthly climatology) and total daily precipitation of the Emilia Romagna region of northern Italy (essentially the Po Valley) by means of multivariate statistical analysis techniques. Firstly, a definition of homogeneous seasons is attempted, based on the analysis of the month-to-month variability of the temporal variance of the parameters, derived from 24 years of daily data (1961-1984). This definition is then adjusted following a monthly empirical orthogonal functions (EOF) analysis of the spatial structures of variability of the same fields. Having defined 'seasons' in this way, attention is concentrated on the cold season, an 'extended winter' ranging from October-November to March-April. Empirical orthogonal functions of both temperature anomalies and precipitation parameters are then recomputed for this extended season. Separate cluster analyses of the low-dimensionality subspaces defined by the first three EOF coefficients (the principal components, PC) of the two parameters reveals the existence of winter weather 'regimes'. Projections of these 'regimes' on geopotential height-fields analyses in the middle (500 hPa) and low (mean sea-level pressure) troposphere are qualitatively consistent with synoptic knowledge of the typical 'weather types' (grosswetterlagen) of the Mediterranean area. A comparison between the cluster analyses of temperature and precipitation shows that mutually compatible thermal and pluviometric regimes can be defined, providing a unified description of the space-time mesoscale variability of the observed parameters.
