An objective definition of seasons for the Mediterranean region is performed using the mean intra-annual variations of 12 climatological parameters. The aim is to achieve a climatologically appropriate determination of the onset and cessation dates, as well as the duration, of each season, better than the conventional seasons (winter, spring, summer and autumn). Moreover, possible changes of the defined seasons during the 70-year study period 1949-2018 are investigated over the climatologically sensitive Mediterranean region. The data used are daily NCEP/NCAR Reanalysis grid point values of precipitation rate, convective precipitation rate, 2 m temperature, total cloud cover, 2 m zonal and meridional wind, 500 and 1,000 hPa geopotential height, 500 and 850 hPa temperature, 850 hPa specific humidity and precipitable water over the Mediterranean region, for the period 1949-2018. Firstly, Principal Component Analysis is applied to the mean intra-annual variations of the above parameters in order to reduce the dimensionality and then k-means Cluster Analysis is applied to the resultant Principal Components in order to group dates with similar regimes of the above parameters. This methodology is applied for the 70-year period 1949-2018 as well as for the five overlapping 30-year subperiods

No abstract

No abstract

Α new method, based on the definition of weather types (WTs), for the determination of seasons in the Mediterranean region is adopted. This method combines principal component analysis and k‐means Cluster Analysis on daily grid point meteorological data for the 70‐year period 1949–2018 obtained from the NCEP/NCAR database. At first, the method is applied to the classification of WTs and subsequently to the definition of seasons based on the mean intra‐annual variations of the WTs' frequency. The resulting seasons are compared to a previous method of the seasons' definition, based on the long‐term mean intra‐annual variations of climatological parameters. Furthermore, a synthesis of the two methods is proposed. The analysis is applied not only to the 70‐year period 1949–2018, but also to the five overlapping 30‐year sub‐periods 1949–1978, 1959–1988, 1969–1998, 1979–2008, 1989–2018 in order to investigate possible long‐term changes of the seasons' characteristics. Four seasons are identified for all periods and for both methodological approaches. These seasons broadly correspond to the four conventional ones, but they differ in their onset and cessation dates as well as their duration. For the composite method, it is found that winter lasts about 4 months, summer is a little longer than 3 months and spring and autumn last around 2.5 months. The most remarkable findings regarding the seasons' characteristics are: (a) the start/end dates and duration of winter are almost identical in all three approaches, while summers' characteristics differ; (b) shortening of winter and spring in recent periods, associated with the delayed start and expedited end dates, respectively; and (c) the prolongation of autumn and summer, due to the delay of its end date, and the earlier onset and delayed end dates, respectively.

No abstract