Towards Consistency in Forecasting Severe Weather Events across a Wide Range of Temporal and Spatial Scales in the FACETs Paradigm

One of the challenges of the FACETs initiative is to provide background information for probabilistic forecasts of hazardous weather that is meaningful over broad time and space scales. Beginning with these larger scales, severe weather reports from the Storm Prediction Center’s report archive were split into 24-hour convective outlook days (12 UTC – 12 UTC) and then analyzed temporally to understand the time scales that hazardous weather events are contained within. Various temporal and spatial scales were tested to find the fraction of reports that are contained within different combinations of those scales. Results indicate that on any given point on a given day, more than 95% of reports within 40 km of the point were captured in a 4 hour period across the United States. Therefore, the SPC’s 24 hour convective outlook probabilities can be interpreted as 4 hour convective outlook probabilities. To test this concept, an experiment was carried out to test how well humans can predict when this 4 hour period will occur for a multitude of severe weather events. Forecasters during the 2016 Hazardous Weather Testbed were asked to predict the 4 hour period in which 95% of severe reports would occur after completing the probabilistic forecast for the severe weather threats of that day. Verification of these timing forecasts reveal that the start time for severe weather events is well forecasted, but event progression is often times too slow. These results begin to address the implications of forecasting hazardous weather events on dynamic combinations of temporal and spatial scales.

Finally, as background to help evaluate the quality of forecasts and provide “practically perfect” forecasting bounds, a local hourly climatology of severe weather probabilities was developed following the methods described in Brooks et al. (2003). These estimates begin the process of analyzing hazardous weather frequencies on a sub-24 hour scale. This work provides a focus on scales broader than the typical time and space scales of the events of interest (including current products on the “watch” scale). A large challenge remains to continue the description as the time and space scales of the forecast become comparable to the scale of the event.