An Assessment of Automated Boundary and Front Detection to Support Convective Initiation Forecasts

One of the largest sources of error in the current automated convective weather forecast systems is due to its inability to accurately account for new convective storm development. In many situations the initiation of new convection is preceded by low altitude convergence in the horizontal winds. These regions of low altitude convergence, often referred to as boundaries, are typically associated with synoptic scale fronts, drylines, and thunderstorm outflows. Gridded wind analyses that utilize Doppler weather radar, surface, and aircraft measurements are one of the best sources of low altitude winds that can be used to identify wind boundaries over large domains.

This study summarizes the preliminary results of a study which examined the feasibility of using gridded wind analyses from operational wind analysis systems to make automated detections of wind boundaries. The analysis focused on two operational wind analysis systems both capable of producing high update, and high spatial resolution wind analyses over a domain that covers the eastern half of the Continental United Sates (CONUS), the Space Time Mesoscale Analysis System (STMAS) and the Corridor Boundary layer wind analysis system (CBOUND). Wind analyses from both systems were first processed with a Lagrangian temporal filter and then passed through an automated boundary detection algorithm based on the Terminal Doppler Weather Radar (TDWR) Machine Intelligent Gust Front Algorithm (MIGFA). The results indicate that the temporal filter improves the boundary signal to noise ratio such that it is technically feasible to make fully automated boundary detections with image processing techniques.