Font Size: a A A

An objective, statistical system for short-term probabilistic forecasts of thunderstorms

Posted on:2003-03-24Degree:Ph.DType:Thesis
University:The Pennsylvania State UniversityCandidate:Hilliker, Joby LeeFull Text:PDF
GTID:2462390011983494Subject:Physics
Abstract/Summary:
A major dilemma facing the aviation industry early in the 21st century is the consequence of high consumer demand for air travel. With passenger numbers at record numbers, the United States air-traffic system has become strained. Inclement weather, such as low clouds, reduced visibility, and thunderstorms, compounds the problem.; Currently, the aviation industry does not have a nationwide system for forecasting high-impact weather. The reason is that the guidance needed by the aviation industry is much different that the guidance that has traditionally been disseminated. Forecasts need to made for the short-term (<6 h), be timely, and have rapid-update capabilities. Moreover, the guidance needs to take into account the inherent uncertainty in weather prediction. Hence, the requirement for objective and reliable probabilistic guidance.; This thesis presents a prototype forecast system that possess these attributes. Specifically, a system is constructed to generate short-term forecasts of thunderstorms. Thunderstorms are chosen because they are the greatest contributing cause of air-traffic delays. In fact, the aviation industry spends {dollar}3 billion a year from thunderstorm impacts.; The system makes uses of archives of high-resolution (both in time and space) observational datasets. These observations are: (a) WSR-88D radar data, (b) profiler data, and (c) surface data from the Oklahoma mesonetwork. After the datasets have been analyzed for bad/missing data, and then replaced with suitable estimates, the most powerful thunderstorm predictors are ascertained.; Results show that radar data have the greatest contribution to skill. There is an increasing contribution from surface, then upper-air data, for longer lead times. The upstream percent areal coverage of reflectivities is the most powerful predictor of thunderstorms for all lead times. The absolute convergence and climatological departure of relative humidity are the most powerful predictors from the surface data. From upper-air data, the most popular predictors are relative humidity and stability observations from the closest radiosonde site.; The system has skill scores of 0.09–0.39 relative to persistence climatology (alternatively, a 9–39% improvement in MSE) through a 360-min lead time. Its performance is highest for lead times ≤30 min, with a gradually drop thereafter. Yet, even for the 360-min lead time, forecast performance relative to persistence climatology remains superior to a statistically significant degree.
Keywords/Search Tags:System, Aviation industry, Thunderstorms, Lead, Forecasts, Short-term, Relative, Data
Related items