Characteristics of a Future Severe Weather Warning Paradigm for FACETs in the
Christopher Karstens, Cooperative Institute for Mesoscale Meteorological Studies (CIMMS) and National Severe Storms Laboratory (NSSL), Norman, OK
Alan Gerard, National Severe Storms Laboratory (NSSL)
Lans Rothfusz, National Severe Storms Laboratory (NSSL)
The primary goal of Forecasting a Continuum of Environmental Threats (FACETs) is to evolve the current National Weather Service (NWS) deterministic watch/warning paradigm to one based on Probabilistic Hazard Information (PHI). Four years of iterative research, development, and experimentation with NWS forecasters, Emergency Managers (EMs), and Broadcasters has provided soft, yet tangible insights into the defining characteristics of this future paradigm for severe convective weather events. This presentation will articulate these findings.
First, a future warning paradigm likely needs to retain deterministic products, but with greater temporal and spatial precision, as well as the inclusion of sub-severe deterministic information (i.e., advisories). Today's forecasters are a byproduct of decades of warning decision-making and training, with demonstrated skill and improvement over time. Emergency manager (EM) participants expressed deep reliance these expert yes/no decisions for justifying decisions of their own, and they have been appreciative of our experimental efforts to provide greater warning precision for individual hazards.
Second, the underlying driver of the current warning paradigm is hazard severity for defining thresholds and estimating intensity. However, predictions of hazard severity are necessarily uncertain. Therefore, a future warning paradigm likely needs to make use of probabilistic information to denote forecaster confidence as a function of lead-time. Geographically, this uncertainty information can be mapped as a plume, giving a relative indication of the area(s) most likely to encounter the severe weather contained within a deterministic product.
Third, the most savvy EMs not only have developed precautionary plans of action for severe weather, but they also have estimates for how long it will take to execute their plans. Thus, a future warning paradigm likely needs to include hazard specific timing information that can be derived and rapidly updated for specific locations. Lead-time is traditionally calculated as the time difference between warning issuance and the first report within the warning. Accounting for time needed to execute a precautionary plan bifurcates this lead-time calculation into 1) the amount of time preceding the decision to execute the plan, and 2) the amount of time needed to execute the plan. The first of these defines the user's specific lead-time.
To achieve these three elements of a future warning paradigm experimentally, we have applied the concept of object identification and tracking, both manually and in a human/machine mix where mechanical aspects of the object updates are partially automated. This allows the forecast to keep pace with the dominant scales of motion associated with severe convection and frees the forecaster to engage in rapidly updating the communicative aspects of a forecast (e.g., intensity changes) when necessary. Unification of the aforementioned warning paradigm elements with forecasters and objects in a real-time severe weather event is a vital process we have termed "continuous flow of information".
This presentation will conclude with thoughts on future directions of research. It is clear that more work is needed to better understand how to establish forecaster trust with sophisticated artificial intelligence decision aids. Additionally, efforts are underway to begin understanding how to bridge the information gap between watches and warnings using probabilistic hazard information.