Abstract
The Local Objective Guidance for Predicting Precipitation Type (LOG/PT) consists of regression equations and nomograms. LOG/PT was designed to address problems inherent in forecasting wintry precipitation across North Carolina, where frozen and freezing precipitation are relatively infrequent, often occurring from mixed precipitation events, and where even small amounts can disrupt communities. Moreover, LOG/PT is an example of employing developmental strategies to maximize the yield from limited resources to produce an objective forecast tool for a critical local-forecast problem.
Stepwise linear regression, with modifications to approximate the sigmoid curve associated with logit regression, was used to derive relationships between precipitation type and 1000–700-, 850–700-, and 1000–850-mb thickness values from radiosonde observations (raobs). The soundings were concurrent with, or within 12 h prior to, the onset of the precipitation at the prediction sites.
The regression portion of LOG/PT discriminates frozen from liquid precipitation. LOG/PT demonstrated skill in detecting frozen events and in correctly specifying frozen-precipitation forecasts. When used in a perfect prog sense with the nested grid model (NGM) thickness forecasts, LOG/PT showed a tendency to overforecast the frequency of snow. LOG/PT's forecast success was limited by its dependence upon a one-raob prediction site with raobs taken 12 h part, and the characteristics of the NGM 1000–850-mb thickness forecasts. Operationally, the regression portion has been useful in predicting the location of the snow/rain boundary in storms with relatively narrow precipitation-type transition zones. In addition, nomograms were prepared to differentiate mixed-precipitation events that resulted in measurable amounts of frozen precipitation from those producing only a trace of frozen precipitation, and to identify icing events. Operationally, the nomograms are used to specify precipitation type in storms with broad bands of mixed precipitation.
In addition to statistical samples, the operational experience of local forecasters was used to gain insight concerning the forecast performance of LOG/PT and the Model Output Statistics (MOS) Probability of Precipitation Type (PoPT) guidance from the Limited-Area Fine Mesh (LFM) model. LOG/PT provides the forecaster with an additional source of objective precipitation-type guidance that can be helpful, especially when forecast errors in the LFM limit the accuracy of the resulting MOS guidance.
Future research efforts directed toward improving the LOG/PT guidance, and increasing the forecaster's knowledge of synoptic features and physical processes that determine precipitation type are also discussed.
