Radok, U., and T. J. Brown, 1993: Anomaly correlation and an alternative: Partial correlation. Mon. Wea. Rev., 121, 1269-1271.


The "anomaly correlation" between a predicted time series and a verifying series, both of which have had the same climatic history subtracted, combines their "direct" correlation with their two climate correlations and with the variance of the historical series. The direct correlation can be freed of climatic effects to become the "partial" correlation, which therefore seems a better parameter for judging predictive skill.