Is Eurasian snow cover in October a reliable statistical predictor for the wintertime climate on the Iberian Peninsula?
Article first published online: 1 AUG 2013
© 2013 Royal Meteorological Society
International Journal of Climatology
Volume 34, Issue 5, pages 1615–1627, April 2014
How to Cite
Brands, S., Herrera, S. and Gutiérrez, J.M. (2014), Is Eurasian snow cover in October a reliable statistical predictor for the wintertime climate on the Iberian Peninsula?. Int. J. Climatol., 34: 1615–1627. doi: 10.1002/joc.3788
- Issue published online: 3 APR 2014
- Article first published online: 1 AUG 2013
- Manuscript Accepted: 23 JUN 2013
- Manuscript Revised: 13 MAY 2013
- Manuscript Received: 10 JAN 2013
- seasonal forecasting;
- statistical forecasting;
- snow cover;
- climate variability;
- Iberian Peninsula
In this study, the recently found lead–lag relationship between Eurasian snow cover increase in October and wintertime precipitation totals on the Iberian Peninsula is re-visited and generalized to a broad range of atmospheric variables on the synoptic and local scale. To this aim, a robust (resistant to outliers) method for calculating the index value for Eurasian snow cover increase in October is proposed. This ‘Robust Snow Advance Index’ (RSAI) is positively correlated with the wintertime (DJF) frequency of cyclonic and westerly flow circulation types over the Iberian Peninsula, while the corresponding relationship with anticyclonic and easterly flow types is negative. For both cases, an explained variance of approximately 60% indicates a strong and highly significant statistical link on the synoptic scale.
Consistent with these findings, it is then shown that the lead–lag relationship equally holds for the DJF-mean conditions of (1) precipitation amount, (2) diurnal temperature range, (3) sun hours, (4) cloud cover and (5) wind speed on the local scale. To assess if these target variables can be skillfully hindcast, simple linear regression is applied as a statistical forecasting method, using the October RSAI as the only predictor variable. One-year out cross-validation yields locally significant hindcast correlations of up to approximately 0.8, obtaining field significance for any of the five target variables mentioned above. The validity for a wide range of atmospheric variables and the consistency of the local- and synoptic-scale results affirm the question posed in the title.