Recently, the National Weather Service (NWS) improved warning services for severe convective weather (hail, wind, and tornado) by transitioning from county-based to polygon-based warning services and verification. This improvement has led to more geographically specific warnings that are intended to focus the area of threat over using geopolitical boundaries. In the present polygon warning method, forecasters 1) make a determination of the geospatial area that will be threatened by severe weather over the course of the warning issuance period, 2) sometimes adjust locations and times of arrival in the path, and 3) update the warning polygon when the storm is about to exit the warned region or when the warning is about to expire (many NWS offices also issue occasional Severe Weather Statements as intermediate updates to warnings and their polygons).

While this polygon approach is a significant step toward the improvement of severe weather warning services, further improvements are possible. Although the warning areas are being refined, the geopolitical boundaries are sometimes still apparent in the polygon shapes. Second, variable updates to polygons lead to a range of lead times for any one location in a polygon. This is especially noticeable for longer-lived multi-warned storms – points in the downstream portions of new warning polygons have increased lead time relative to the upstream portions. Also, the likely time of arrival and departure of expected severe weather areas are particularly challenging to convey within the polygon.

In this study we evaluate a new approach to creating warning areas in which the forecaster instead outlines the probable severe weather threat area at the current time, and using user-supplied motion estimates and uncertainties, creates a high temporal and spatial resolution geospatial warning grid. This grid can be used to generate a wide variety of useful warning information, including warning swaths over a time period, meaningful information for time of arrival and departure for any location within the warned area, and probabilistic information based on the storm motion uncertainty. This method easily lends itself to the inclusion of forecaster uncertainty trend information over the time period of the warning, leading to a true probabilistic gridded warning product, which is a potentially significant evolutionary step in warning improvement.

We will collect a number of actual warned storms from a variety of locations nationwide and with a variety of storm types. At the issuance time of these warnings, the polygon swaths will be augmented by a reasonable depiction of the severe weather threat area. These threat areas will then be integrated over time, using the storm speed and direction and a reasonable estimate of motion uncertainty to produce continually-evolving threat area and swath grids. We will report statistics on the improvements to warning services at all points within both the threat area swaths using our method and the original warning polygons, including improvements to lead time, and improvements to probabilities of detection, false alarm rate, and false alarm area.