Search

A statistical system that uses surface observations and radar data to provide 1-, 3-, and 6-h forecasts of temperature, dewpoint depression, and wind speed is developed. Application of the system to independent data demonstrates that the radar data result in a 3%–6% reduction in the mean-squared error (MSE) of the temperature forecasts when compared with that when only surface observations are used as predictors. The addition of the radar data similarly provides a 2%–4% reduction in MSE of the dewpoint depression forecasts. However, there is no difference in skill when the radar data are included in forecasts of wind speed. The majority of the radar-data predictors selected by the stepwise regression algorithm for inclusion in the forecast system are from areas that are poorly sampled by surface observations. This result supports the argument that more uniform sampling (better coverage) of the environment is playing a role in the improvement of the forecasts. Moreover, analysis of individual even...

An automated statistical system that utilizes regional high-density surface observations to forecast low ceiling and visibility events in the upper Midwest is presented. The system is based solely upon surface observations as predictors, featuring forecast lead times of 1, 3, and 6 h. A test of the forecast system on a 5-yr independent sample of events shows that for a 1-h lead time, an additional 2%–4% reduction in the mean squared error (MSE) is obtained by the high-density forecasting system compared to that for a system utilizing only the standard synoptic observations. Meanwhile, tests on a 3-h lead time reveal an additional 0%–1.5% reduction in MSE by the high-density system over the synoptic system. Little improvement is gained by the high-density system at a 6-h lead time. The results indicate that current observations-based forecasting techniques can be improved simply by utilizing a higher density of surface weather observations. With this enhanced guidance, it is likely that decisions ...

Several methods of generating very short term (0–6 h) probabilistic forecasts of ceiling and visibility are investigated: 1) an observations-based (OBS-based) system in which potential predictors consist of weather observations from a network of surface stations along with several climatic terms; 2) the traditional model output statistics (MOS)-based approach in which potential predictors consist of nested grid model (NGM) output, the latest observation from the forecast site, and climatic variables; and 3) persistence climatology in which potential predictors consist of the latest observation of the predictand variable from the forecast site and several climatic terms. Forecasts are generated for each technique on 2 yr (1993–94) of independent data for 25 stations in the eastern United States. Two variables (ceiling and visibility) are forecasted for eight thresholds, two initial times (0300 and 1500 UTC), and three lead times (1, 3, and 6 h). Results show that the OBS-based method is superior t...

The skill of probabilistic Model Output Statistics forecasts generated from Generalized Additive Models (GAM) is compared to that of traditional multiple linear regression techniques. Unlike linear regression, where each predictor term in the additive model is assumed to vary linearly with the predictand (unless specified otherwise by the developer), GAM is a nonparametric tool that makes use of the data to automatically estimate the appropriate functional (curvative) relationship for each predictor term. This relieves the developer from the chore of identifying and computing the correct predictor transformations and helps uncover certain nonlinearities that may have been missed. Forecast equations for each statistical technique are developed for nine regions encompassing a total of 90 stations in the northeastern United States. Three parameters (cloud amount, ceiling height, and visibility) are forecast for eight thresholds and two lead times (12 h and 24 h). The developmental dataset consists o...

The skill of the limited-area fine-mesh (LFM) model in making 24-h areal quantitative precipitation forecasts (QPF) of ⩾ 0.5 in. (12.7 mm) is evaluated and analyzed for two warm seasons; 1982 and 1983. Differences in skill between the eastern and western United States are investigated and the impact of updated initial conditions is explored. The skill in predicting precipitation from cyclonic versus mesoscale systems is also examined. It is found that the model skill changed drastically from the summers of 1982 to 1983. Skill levels for 1982 were about 25% to 35%. In 1983, they dropped to an average value of less than 10% and often to zero on a daily basis. The precipitous drop in skill appears to be the result of changes made in key model threshold parameters. These parameters include grid-resolvable saturation criterion and convective cloud base boost. The introduction of constraints on convective precipitation and the vertical advection of moisture also appear to have reduced model skill. A co...

Satellite, radar, surface, and upper-air data from the June–August periods of 1982–83 are examined to determine the mechanisms, and their relative contributions, for producing warm-season precipitation in the United States. Only areas where rainfall equaled or exceeded 12.7 mm during the 24-h period ending at 1200 UTC were considered. Rainfall associated with extratropical cyclones accounted for about half of the warm-season precipitation. Most of the remaining precipitation was produced by mesoscale forcing mechanisms acting independently from the traveling extratropical cyclones. Nearly all the warm-season precipitation was convective; over 80% of the total was directly or indirectly associated with thunderstorms. Nearly three-fourths of the precipitation that occurred between the Rockies and the Mississippi Valley was nocturnal. Conversely, about three-fourths of the precipitation east of the Mississippi Valley fell during the daylight hours. During the 1983 July-August drought, the area of pr...