Persistence of U.S. Seasonal Temperatures up to One Year

Abstract

An attempt is made to physically explain the distribution of lag correlations (with long-period trend removed) from one to four seasons based on 40 years of data at about 200 stations in the conterminous United States. These data are analysed with the help of histograms (and medians) for climatologically homogeneous areas and a series of isopleth charts. The results indicate that seasonal persistence is most pronounced 1) when summer is antecedent, 2) in most seasons along the West and East Coasts and near the Great Lakes, and 3) in areas and times where and when variability of snow cover or antecedent precipitation is large. Negative persistence occurs mainly between successive Springs and at certain seasons over the Western Plateau. Physical concepts involving heat storage and teleconnections from large-scale coupled air-sea systems over the North Pacific are suggested as causes. Unfortunately, the correlations are for the most part too small to be used as a sole guide for making long-range forecasts.