An Object-based Approach For Identifying and Evaluating Convective Initiation

Abstract

Managing the safe flow of aviation air traffic in hazardous weather conditions is dependent upon weather forecasts that accurately depict the initiation of convective thunderstorms 4, 6 and 8 hours in the future. Since forecasts of convective initiation are in their infancy, they require improvement to become a reliable and accurate depiction of the hazardous weather for improved flight planning. In order to measure the accuracy of the convective initiation element of a forecast, new verification techniques are being developed and tested. One specific technique used to assess the National Center for Atmospheric Research's AutoNowcaster is the Procrustes Technique, an object-based verification approach. In general, the Procrustes object-based verification scheme identifies objects in the observation field and compares them to the similarly defined forecast objects. The verification scheme matches like objects and assigns a penalty based on minimizing the cost function of all possible matches. In the end, the unmatched objects are considered initiation if certain requirements are met. The use of this object-based technique will be presented in the context of assessing convective initiation forecasts for use in aviation air traffic planning.