Comment on "Winter 'weekend effect' in southern Europe and its connection with periodicities in atmospheric dynamics" by A. Sanchez-Lorenzo et al

Abstract

Weekly periodicity of winter pair data of 12 Spanish meteorological stations for the period 1961-2004 was investigated. Contrary to SL08, no significant weekly cycle in the ppir data was found. MC analysis of the Spanish data shows that two-third of stochastic realizations have larger weekly amplitudes of pair deviations from the overall mean than the weekly amplitude of the measurement data. Also the shape of the weekly cycle can be explained by randomness, as the MC analysis and additional statistical testing showed. The results from SL08 can be reproduced if p air time series are assumed to be independent. However, all the different time series of pair data are strongly correlated. SL08 also found that one of nine European atmospheric circulation patterns differed significantly between different weekdays. However, the fact that only one out of nine circulation patterns shows a significance at the 95% level is itself not significant. Therefore, the supposed statistically significant weekly cycle of meteorological variables is not significant (and therefore not proven) and the result of a statistical analysis that neglected the statistical dependency between pair time series. However, this does not mean that such a weekly cycle would not exist. Bäumer and Vogel [2007] found a weekly cycle in German meteorological data, but the significance of that cycle for precipitation and sunshine duration was questioned [Hendricks Franssen, 2008; Barmet et al, 2009]. Laux and Kunstmann [2008] concluded that temperature exhibits a significant weekly cycle in Europe, whereas precipitation and sunshine duration do not. Copyright 2009 by die American Geophysical Union.