A Statistical Approach to Short-Term Thunderstorm Outlooks

View More View Less
  • 1 Techniques Development Unit, National Severe Storms Forecast Center, Kansas City, MO 64106
© Get Permissions Rent on DeepDyve
Restricted access

Abstract

A multiple discriminant analysis program is used to obtain prediction functions for general and severe thunderstorm activity during April and July. Four predictors, lifted index, mean low-level mixing ratio, K index and mean 200–300 mb divergence, are statistically combined for an area east of the Rockies. Forecasts are valid for a 12 h period. Tests on dependent and independent data using a probability of detection (POD), a false alarm rate (FAR), and a critical success index show stability for the prediction functions. Operational use of the prediction functions is examined from two approaches, one involving POD and FAR and the other involving a more conventional probabilty approach. Semi-operational results from April and July 1979 evaluations show statistical successes comparable to the dependent data results.

Abstract

A multiple discriminant analysis program is used to obtain prediction functions for general and severe thunderstorm activity during April and July. Four predictors, lifted index, mean low-level mixing ratio, K index and mean 200–300 mb divergence, are statistically combined for an area east of the Rockies. Forecasts are valid for a 12 h period. Tests on dependent and independent data using a probability of detection (POD), a false alarm rate (FAR), and a critical success index show stability for the prediction functions. Operational use of the prediction functions is examined from two approaches, one involving POD and FAR and the other involving a more conventional probabilty approach. Semi-operational results from April and July 1979 evaluations show statistical successes comparable to the dependent data results.

Save