No abstract

The paper reports the main results of the EU project Millennium in the Mediterranean area over the last 500 years. It analyses a long series of temperature from Portugal, Spain, France, Italy and Greece. The series are obtained by combining indices from documentary sources from AD 1500 to the onset of regular instrumental observations. There is an ongoing discussion regarding the proper way of combining documentary and instrumental data and how to translate accurately the conventional indices from −3 to +3 into modern units, i.e. degree Celsius. This paper produces for the first time a number of early instrumental observations, in some cases (i.e. Italy and France) covering 350 years, including thereby the earliest regular observations after the invention of the thermometer. These Mediterranean data show that anomalous temperatures usually had only a locally limited effect, while only few extreme events had a widespread impact over the whole region, such as the summer of 2003. During the period from 1850 to the present day, the Mediterranean temperature anomaly was close to the Northern Hemisphere in spring and summer, while it was warmer in autumn and winter. Compared with the long-term instrumental records (i.e. 1655 onwards), the recent warming has not exceeded the natural past variability characterized by heating-cooling cycles with no significant long-term trends.

Abstract. Day-to-day temperature variability is investigated in eight European series of daily mean temperatures beginning between 1722 and 183 3. Eight statistical measures of day-today temperature variability are compared. The intramonthly standard deviation of daily temperature anomalies is found to be a good measure. The absolute change in temperature anomaly from one day to the next is sensitive to changes in observational procedures and is suggested as a diagnostic tool for identification of inhomogeneities in instrumental temperature series. Because many changes in observational procedures have taken place, quantitative estimates of trends in day-to-day variability, based on all series, could only be calculated for 1880-1998. A trend analysis over this period indicates an increase by 5% in southwest Europe, 0 to -5% change in the northwest, and a decrease by 5 to 10% in northeast Europe. On a longer time perspective, day-to-day temperature variability in winter, spring, and autumn in northern Europe has decreased over the last 200-250 years. The frequency of extremely cold •vinter days in northern Europe •vas lower in the twentieth century than in the eighteenth and nineteenth centuries. Day-to-day temperature variability in •vinter in northern Europe was negatively correlated with a North Atlantic Oscillation index in the period 1826-1997, but some other factor must also have contributed to the long-term variability decrease. More long daily temperature series, and development ofhomogenization methods for such data, are needed for an improved knowledge of long-term changes in day-to-day temperature variability.

No abstract

No abstract