The Impact of High-Frequency Surface Weather Observations on Short-Term Probabilistic Forecasts of Ceiling and Visibility

Abstract

The skill of an automated statistical forecasting system that uses only hourly (top of the hour) surface observations is compared with a system that utilizes hourly and high-frequency (interhour) surface observations to forecast low-ceiling and low-visibility events in the New York City, New York, area. Forecasts feature lead times of 1 h and less. Equations to forecast ceiling and visibility conditions 1 h in advance are developed for initialization times at the top of the hour, as well as at 15, 30, and 45 min past the hour. Two forecasting systems are created: a baseline system that utilizes only hourly surface observations and an alternative system that utilizes both hourly surface observations and high-frequency observations. Introduction of the high-frequency observations into the forecasting system produces an additional 1.5%?4.5% reduction in the mean-square error (MSE), as compared with the baseline system for 1-h forecasts made at the top of the hour. By 45 min past the hour, the reduction in MSE over the baseline system increases to 14%?17%. The high-frequency observations are also utilized to develop forecast equations with lead times of 5?55 min. Reduction in MSE for this rapid-update forecasting system in comparison with simple persistence increased from an average of 3% for a 5-min lead time to an average of nearly 22% for a 55-min lead time. Moreover, improvements over persistence climatology increased from an average of 1.5% for a 5-min lead time to an average of about 14% for a 55-min lead time. These findings indicate that current observations-based forecasting techniques can be improved by utilizing high-frequency surface weather observations. Therefore, the uncertainty in decisions affected by the arrival and duration of a low ceiling and low visibility can be reduced, thereby providing enhanced guidance for operational airport traffic delay programs.