An Empirical Correction to the Barotropic Forecast

An investigation of error fields in 30 consecutive barotropic forecasts for the 500 mb surface has been performed. It is found that the error fields show persistence in the time to such an extent that a correction consisting of the mean error field for a number of preceding forecasts can be applied to a barotropic forecast with advantage. The improvement of the forecast depends upon the number of preceding forecasts included in the mean error field. The greatest increase in accuracy is found when 10–15 preceding forecasts are contributing to the correction field. The 24 hour forecasts are improved to the greatest extent by the correction, but the 48 hour-forecasts also show a significant increase in accuracy. The same is true to a smaller extent for the 72 hour-forecasts.

In section 3 an investigation is made as to whether the height change obtained by making barotropic forecasts from the time-averaged map for the period investigated or from the normal 500 mb map for the corresponding month can be used with the same success as the correction used in section 2. It it found that the correction from the mean chart consistently improved the forecasts. The overall effect is, however, small. The correction obtained from the normal chart did not affect the accuracy of the forecasts appreciably.