A vision for the development and implementation of the Warn on Forecast concept
Pamela Heinselman
1 Dec, 10:30 am, in 2155
Abstract:
Deterministic modeling approaches to forecasting severe weather presume
a single solution for the evolution of storms and the environments in
which they form. Similarly, deterministic severe weather warnings
assume severe weather impacts are equally likely within the warning’s
spatial and temporal extent. This approach fails to represent forecast
uncertainty that we know exists, owing to the chaotic nature of the
atmosphere, imperfect observations, imperfect prediction models, and
other causes. Advancements in data assimilation, ensemble modeling,
high-performance-computing, and probabilistic forecast and verification
methods provide the means and opportunity to better represent these
uncertainties within severe weather prediction systems.
The Warn-on-Forecast project at the National Oceanic and Atmospheric
Administration (NOAA) National Severe Storms Laboratory in Norman, OK,
USA, aims to improve severe weather forecasts, warnings, and decision
support for high-impact events (e.g., tornadoes, hail, wind, and
flooding) by leading convective-scale research and development
activities that enable a new paradigm where convection-allowing,
ensemble model forecasts become a key resource for NOAA National
Weather Service (NWS) Watch and Warning operations. The result of these
efforts is the development of a prototype Warn-on-Forecast system
called the NSSL Ensemble Warn-on-Forecast System for ensembles
(NEWS-e).
NEWS-e is a frequently updated, regional-scale, on-demand
convection-permitting ensemble analysis and prediction system, nested
with an hourly convection-allowing ensemble forecast system. The 2017
version of this system assimilates radar, satellite, and surface data
every 15 minutes, and generates new probabilistic 3-hour and 1.5-hour
forecasts at the top and bottom of each hour, respectively, at grid
spacing O(~3 km). This multiscale data assimilation system uses the
advanced research version of WRF, version 3.8+ (ARW) to produce
storm-scale ensemble analyses and forecasts. Details of the system
configuration will be provided within the presentation.
NEWS-e seeks to improve 0–3-h predictions of individual convective
storms and mesoscale aspects of convection that provide enhanced
probabilistic forecast guidance. Forecast swaths produced by this
system, such as probability of simulated reflectivity > 40 dBZ and
ensemble 90th percentile values of accumulated rainfall, 2–5-km updraft
helicity, and 0–2 km vertical vorticity, are expected to revolutionize
forecasters’ ability to anticipate not only storm location, mode,
intensity, but also high-impact threats and their impacts on society.
Toward this end, this prototype system is being tested in real time
during peak severe weather season in the U.S. Following each storm
season, findings from rigorous quantitative and qualitative case study
evaluations are used to direct enhancements of the system. Example
cases and the associated forecast verification will be shown.