Map-Pattern Classification by Statistical Methods

Abstract Sea-level pressure values at 22 northeastern United States stations were used to describe the map pattern for each of 445 winter days. Each map pattern was correlated with every other map pattern in order to determine whether some patterns reappeared more frequently than expected by chance. A number of distinctively different patterns, called types, were isolated. These types are believed to he useful in stratifying map patterns for further study. The weather accompanying most of the types is shown to depart substantially from the climatology of the data sample.