Experience in Statistical Forecasting of the Atmospheric Temperature

Abstract

In this paper, the problem of forecasting of quantitative features of the weather (atmospheric temperature, atmospheric pressure, direction and velocity of wind, relative humidity of the air and some other) is considered. The problem is formulated as a task of linear extrapolating a vector random sequence. Such forecast with statistical estimations of parameters of predictive algorithm was named as linear statistical forecast. Empirical data about precision of linear statistical forecast of the atmospheric temperature in comparison with precision of the official numerical forecast and climatological forecast is given and analyzed. The experience of forecasting shows that statistical forecast is more accurate than climatological forecast and a little less accurate than numerical forecast.