Experiments in Objective Aviation Weather Forecasting Using Upper-Level Steering.
Final rept. 1 Oct 81-30 Sep 83,
AIR FORCE GEOPHYSICS LAB HANSCOM AFB MA
Pagination or Media Count:
This report describes a series of three experiments designed to test the feasibility of using upper-level steering of weather patterns to produce short-range local forecasts. Objective forecast models were developed and tested using fall and winter data from northeastern United States. The complexity of the models was increased in attempts to treat problems related to coexistence of stationary and traveling patterns and the effects of diurnal changes in the boundary layer. The first model simply steers or advects aviation weather parameters analyzed on an x-y grid by looking upstream at 1-hour time steps. The model was tested during a 3-week period using the AFGL McIDAS facility, preparing 0-15-hour forecasts of wind, cloud cover, visibility, and dewpoint at three locations. Two steering flows used were the 700-mb wind field and the vertically integrated 850- to 300-mb wind field. In general, the root-mean-square rms errors were worse than persistence for 0-9 hours. Because errors were introduced when stationary orographic patterns were advected together with traveling systems, the forecast model was revised to advect the change-with-time fields. This model was tested on 12 winter cases. Scores for vector wind, cloud amount, and temperature were slightly better than both persistence and model-output-statistics MOS at 0-3 hours, but were worse at longer periods. One difficulty was that the diurnal changes reversed sign several hours into the forecast. A third model was developed that treats the diurnal changes separately. This model was tested on 12 days in March, producing forecasts of wind, cloud cover, visibility, temperature, and dewpoint at 12 stations.