THE DEVELOPMENT OF EFFICIENT LINEAR STATISTICAL OPERATORS FOR THE PREDICTION OF SEA-LEVEL PRESSURE

Abstract

The design of efficient linear statistical operators for the 24-hour prediction of the sea-level pressure distribution over the United States is considered. Factor analysis techniques for reduction and selection of independent variables in regression analysis are used as a means of obtaining efficient statistical forecasting equations. The effects of the variations in data density in time and space, and the extent of geographical coverage upon the explained variance of the sea-level pressure are examined.