Construction and assessment of a logistic regression model applied to short-term forecasting of thunderstorms in León (Spain)

Abstract

Forecasting thunderstorms is one of the most difficult tasks in weather prediction due to the scarce knowledge on how to characterise the mechanisms taking part in the formation of thunderstorms. New tools based upon the objective recognition of appropriate conceptual models have been developed in the last years. A good example of this are the statistical models, based on variables that characterise the preconvective atmospheric conditions. This paper presents the results obtained by applying a short-term forecast model to thunderstorms during the summer periods in León (Spain). This model makes use of the logistic function as a binary forecasting technique determining storm/no storm. The logistic function used was a combination of 15 previously selected meteorological variables. The model has been constructed with the data collected on 152 occasions, and it has been verified on 77 other occasions. The skill scores obtained show that the use of statistical models combining the data, provided by variables characterizing the preconvective conditions and the triggering mechanisms, represent an important step in the difficult task of short-term thunderstorm forecasting.