Ian Gesell - Feb 28

Convective Meteorology (Mesoscale Dynamics) Seminar

Forecasting a Continuum of Environmental Threats (FACETs): Verification of the Tornado and Lightning Plumes and developing a new Plume Kernel

Ian Gesell

Friday, February 28^th

3:00pm/NWC 5600

Forecasting a Continuum of Environmental Threats (FACETs) is a proposed framework that will modify the current watch and warning system to communicate clear and simple hazardous weather information to serve the public. A key part of FACETs is Probabilistic Hazard Information (PHI). Currently warnings are yes/no (binary). PHI changes this by putting threat probabilities onto grids. Probabilistic plumes were created by forecasters in the NOAA Hazardous Weather Testbed (HWT) as a part of FACETs with the use of the PHI Tool.  These probabilistic plumes are produced using the storm size and speed to determine the width and length of the plume. A Gaussian smoother from the center point is used to provide a visualization of the forecast threat area and region of uncertainty. A method of forecast verification is needed to determine the accuracy of these plumes. Currently the SPC verifies outlooks by a correct forecast corresponds to an event occurring within 25 miles (40 kilometers) of a forecast point. This leads to the question What is an Event? The verification data used for the lightning plumes is the National Lightning Detection Network. The tornado plumes were verified by tracking the mesocyclone coordinates using start and end time from Storm Data. When you are in a warning how to close a tornado do you need to be to consider being in the warning justified? Four cases were taken from the 2017 HWT for the lightning and tornado hazards. For each case, the maximum probability of the plumes at all timesteps for a case were merged together. Attributes diagrams were used for the verification. What radial distance away from an event makes the attributes diagram meaningful? As a result, of the struggle to find meaning using attributes diagrams practically perfect plumes were created for each of the cases. For each of the tornado and lightning cases, it was found that using a 15 km grid around the mesocyclone point was where the attributes diagrams indicated the plumes were the most accurate. It was found that creating practically perfect plumes is a way to help evaluate the reliability of the tornado and lightning plumes. Additionally, different kernels were applied to the forecaster created probabilities to see if there is a better kernel than Gaussian for either of the tornado or lightning plumes. For the tornado plumes there is not a better kernel than Gaussian kernel. However, for lightning there Epanechnikov and/or Triangular are better kernels than the Gaussian kernel.