VLab Forum Members,
This is a reminder that the March VLab Forum is scheduled for Monday,
March 31, at 11:00am EDT. The the topic will be, "
The March VLab Forum is scheduled for Monday, March 31, at 11:00am
EDT. The the topic will be, "Combining guidance from AI/ML using
both observations synthesized through the Probabilistic Hazard
Information application and the Warn-on-Forecast System" (a
modified title from what I sent last week.) The presentation will be
given by Dr. Kristin Calhoun and Dr. Eric D. Loken, who are research
scientists at the NOAA National Severe Storms Laboratory. An
abstract and information about the presenters are included below.
To participate in the VLab Forum, please register
for the GotoWebinar.
To add the VLab Forum session to your calendar, click on this link.
An agenda can be downloaded at this link.
Abstract
The period between watch and warning issuance (corresponding to lead
times of about 1-4 hours) can be important for decision makers who
need longer lead times to take protective actions. However, between
watches and warnings, there are no standardized National Weather
Service (NWS) products designed for public consumption, no guarantees
of consistent messaging between NWS offices, and relatively few
forecaster tools that consider both observations and numerical weather
prediction (NWP) data.
To help address some of these challenges, the National Severe
Storms Laboratory (NSSL) and Cooperative Institute for Severe and
High-Impact Weather Research and Operations (CIWRO) have created
several probabilistic forecasting tools designed to provide guidance
between watches and warnings. Storm-based Probabilistic Hazard
Information (PHI) is designed to provide a meaningful quantification
of hazard likelihood with additional spatial and temporal precision.
Artificial intelligence/machine learning (AI/ML) guidance provides the
initial estimates of hazard probabilities for tornadoes, hail, wind,
and lightning. PHI plumes are continuously updated to reflect
immediate changes in storm motion and intensity. Another AI/ML
product, called Warn-on-Forecast System - Probabilistic Hazard
Information (WoFS-PHI) blends observation-based information from PHI
and NWP forecast data from NSSL’s WoFS to make probabilistic
predictions of severe hail, wind, and tornadoes within specified time
windows and spatial radii at lead times up to 4 hours.
PHI and WoFS-PHI were evaluated in the 2024 Hazardous Weather
Testbed Watch-to-Warning-Experiment (HWT W2WE), which took place over
3 weeks from August - September of 2024. In this experiment, SPC and
WFO forecasters used PHI and WoFS-PHI to issue novel probabilistic
watch-to-warning products in 4 displaced-real-time severe weather
cases. During the final case each week, 3-5 emergency managers (EMs)
used forecasters’ products to make decisions in a parallel EM activity.
In surveys from the W2WE, forecasters indicated that WoFS-PHI
was helpful for creating probabilistic products, especially
non-technical “local discussions” and public graphics. Half of the EMs
indicated that forecaster-issued PHI plumes allowed them to make
decisions earlier than they otherwise would have been able to.
Overall, results suggest that AI/ML products such as PHI and WoFS-PHI
can be used to create more effective messaging to end users, leading
to better and faster decisions.
About the Presenters
-
Dr. Kristin Calhoun has been a research scientist
with NSSL since 2010. Her research focuses on advancing severe
weather forecasting by developing and transitioning new research and
algorithms to NWS operations. She has led multiple experiments in
the NOAA Hazardous Weather Testbed including the development and
testing of Probabilistic Hazard Information, GOES Risk Reduction,
and the integration of lightning data into operations.
-
Dr. Eric Loken received his undergraduate degree
from the University of Wisconsin-Madison and his master's and Ph.D
degrees in meteorology from the University of Oklahoma. Eric has
been working as a research scientist with CIWRO since 2021. The bulk
of his work has been on creating and evaluating products for severe
weather prediction.
