Trends in intraseasonal temperature variability in Europe: comparison of station data with gridded data and reanalyses

Trends in temperature variability are often referred to have higher effect on temperature extremes than trends in the mean. We investigate trends in three complementary measures of intraseasonal temperature variability: (a) standard deviation of mean daily temperature (SD), (b) mean absolute value of day to day temperature change (DTD) and (c) 1-day lagged temporal autocorrelation of temperature (LAG). It is a well-established fact that different types of data (station, gridded, reanalyses) possess different temperature characteristics and particularly its trends. Moreover, it has been uncovered during our research that trends in measures of variability are more sensitive to data inhomogeneities. Therefore, we use five different datasets, one station based (ECA&D), one gridded (EOBS) and three reanalyses (JRA-55, NCEP/NCAR, 20CR), and compare them. The period from 1961 to 2014 where all datasets overlap is examined and the linear regression method is utilized to calculate trends of investigated measures in summer and winter. Intraseasonal temperature variability tends to decrease in winter, especially in eastern and northern Europe, where trends below -7% per decade are detected for all measures. Decreases in DTD and LAG (increase in persistence) prevail also in summer while summer SD tends to increase. The increase in the width of temperature distribution and the simultaneous increase in persistence indicate a tendency towards the rise in the frequency of extended extreme events in summer. Our results do not imply that reanalyses are the least accurate in determining trends of temperature variability. JRA-55 appears to be the least diverging from other datasets, while the largest discrepancies were detected for DTD at climate stations. Our consequent research expose that the type of calculation of mean daily temperature is crucial to the resulting value of DTD as T_{(7+14+21+21)/4} shows warm days warmer and cold days colder than T_(0-0)/24. This is the possible explanation of discrepancies in DTD trends as many of ECA&D stations (e.g. German stations at 2001) underwent a change in this calculation during the observed period.