Post on 27-Dec-2015
Warn on Forecast Briefing September 2014
Warn on Forecast Brief for NCEP planning
NSSL and GSD
September 2014
Warn on Forecast Briefing September 2014
Warn on Forecast Vision (~2020+)• Short-term NWP forecasts at 0.5-1.0 km resolution
• Using improved models for convective-scale NWP
• Incorporation of high-resolution satellite and radar data
• Sub-hourly update cycle with 2-3 hour forecasts
• Providing uncertainty estimates via an ensemble system
• New paradigm for severe weather warnings
• Blend WoF output with observed/historical radar data• Blending using the Multi-Year Reanalysis Of Remotely-Sensed Storms (MYRORSS)
• Similar approach is used by NHC
• Extending lead times (1 hour for tornadoes, 2-3 hours wind, QPF, hail)
• Probabilistic warnings visualized using)
• Probabilistic Hazard Information tools (PHI)
• Forecasting A Continuum of Threats tools (FACETS)
• Leveraging societal knowledge and its technology
• Improving warning effectiveness via social science research
• Communicating threat information via new media
Warn on Forecast Briefing September 2014
• Assimilation• relative value of hybrid versus pure ensemble approaches as well as multi-resolution approaches
• ensemble ICs and mechanisms for maintaining optimal forecast spread (uncertainty)
• dealing with model biases
• use of dual-polarization data
• Observations• data quality control for dBZ, Vr, and dual-polarization variables
• optimal combinations of surface, radar and satellite
• appropriate superobbing for dense observational data
• Models
• Reducing model errors in PBL, radiation, and land surface
• appropriate microphysics relative to observations (e.g., single-pol versus dual-pol radar data)
• Verification• quantitative ensemble object based verification
• best methods to quantify model skill for these types of high-impact & rare events?
• value of high-resolution WoF system for non-severe convective weather
• Predictability
• what is the practical predictability for different environments and weather threats?
• how long does high-resolution satellite/radar data assimilation positively impact forecasts?
• hail and lightning?
Current Research Issues
Warn on Forecast Briefing September 2014
• Operational Implementation?• Background: Hourly-updated ~60 member 3 km convective-permitting model
• Warn on Forecast system:
• 0.5-1 km resolution with ~60 members
• Regional grid (1500-2000 km size grid, possibly moving)
• 5-15 min cycling with 1-3 hour forecasts generated 4-6x per hour
• 15 min latency (e.g., forecast is available T+15 minutes).
• Inputs: sub-hourly radar, satellite (GOES-R?), mesonet obs.
• Outputs: rotation tracks and intensities, QPF, severe wind probabilities, hail, etc.
• Operational Use?• WoF system is an ON-DEMAND service, much like HWRF is today.
• NOAA Storm Prediction Center would control placement and activation of WoF system (that would be run at NCEP)
• Output would be made available to SPC and WFOs through products
• WFOs might “blend” products locally for their needs with other observations(?)
• FACETS and PHI tools would consolidate model output and observations for forecasters into probabilistic threat information
Operational Warn on Forecast?
Warn on Forecast Briefing September 2014
• Research and development of a Warn on Forecast system is on track for end-of-decade specification for NCEP
• NSSL and GSD are working closely together to develop the high-resolution and high-frequency NWP system to provide probabilistic hazardous weather information for watches and warnings in 2020s.
• Given current state of NWP science for convective scale:
• high-resolution & high-frequency will likely regional grid initially (~40K cores?)
• system runs on-demand at NCEP? (like HWRF does?)
• SPC: logical center to decide on location for WoF on high-impact days.
• SPC and WFO’s can collaborate as to configuration (similar to how the convective outlooks and watches are already done)
• unclear yet whether sub-hourly 0.5-1 km forecasts will be useful for non-severe convective weather situations.
Summary