Persistence of observed air temperatures in Iceland

Abstract

A large data set from 40 weather stations in Iceland is explored for persistence in monthly mean temperatures. There are great seasonal and regional variations in the persistence. Extremely high values of correlation (r > 0.8) of temperatures with subsequent months are found. These values are higher than reported elsewhere in the scientific literature. The highest values are found in coastal regions in the summer, while in the early winter there is overall little correlation. In general, there are two distinct maxima in the temperature correlations, one in late winter/early spring and one in the summer. In most seasons, there is greater correlation at the coast than inland. The high correlations are linked to snow melt, persistence in sea surface temperatures, weak winds and strong static stability. Remarkably low correlations reveal a negative feedback process: A warm May leads to less snow in inland regions, which favours a cold sea breeze in coastal regions in June. In regions of high correlation, the persistence is indeed useful for subseasonal forecasting of mean temperatures in late winter/spring and summer.