On the use of time series concepts and spectral and cross-spectral analyses in the study of long-range forecasting problems
Monthly values of sea surface temperature anomalies and wind anomalies for an area of the North Atlantic Ocean are studied as a vector Gaussian process. The covariances and the spectra and cross-spectra provide information on the nature of the interaction of ocean and atmosphere. The long period fluctuations are found to be fairly coherent. Indications are that the sea surface temperature can be predicted for one month into the future on the basis of past values with a 50 % reduction of variance. Prediction problems are discussed.
Pierson, Willard J.. 1960. "On the use of time series concepts and spectral and cross-spectral analyses in the study of long-range forecasting problems." Journal of Marine Research 18, (2). https://elischolar.library.yale.edu/journal_of_marine_research/950