Transcript Societal Impacts of Severe Weather – The Future of

Societal Impacts of Severe Weather – The Future of
Prediction, Communication and Post-Event Analysis
Neil A. Stuart
NOAA/NWS Albany, NY
NROW X
5 November 2008
Increasing importance to evaluation
of societal impacts
•
•
•
•
•
Urbanization of America/World
Increasing population density
Increasing exposure to hazardous weather
Huge diversity in user community
Huge spectrum in levels of vulnerability in
user community
• Increased liability for economic impact and
loss of life
Current system of pre-event and postevent evaluation of forecast value
• Sources of guidance contribute to forecast confidence of
forecast scenario
• Individual forecaster perception of probability of event
initiates forecast/watch/warning
• Graphics and text products convey level of urgency
depending on perceived potential impact
• Observed weather – calculate statistics for POD, FAR,
CSI
• Future trends – Ensemble guidance provides more
quantitative probabilities for various hazards
• Impacts from various hazard types evaluated
• Create text and graphical products (Including PQPF?)
conveying potential impact of various hazards
Valentine’s Day 2007
•
Despite advances in data analysis,
assimilation and visualization,
some important user groups are not
benefiting
–
–
–
–
–
•
•
•
•
•
Widespread 20-42” of snow Capital
Region of NY and north and west
NESIS Category 3 – ranked near
Blizzard of ’78 in SE New England
I-80 shut down in PA due to
accidents in mixed precipitation
Many planes stranded on runways
for hours at JFK airport
35 deaths
Winter Weather Impact checklist
developed at NWS Buffalo for Lake
Effect events
Distributed across eastern region of
NWS
Evaluation of potential utility with
synoptic-scale snowstorms
Ranking of multiple types of
impacts
Accumulated rank defines High,
Moderate or Low Impact
Interior New York/New England
•Timing – 3: Covered multiple rush hours
•Seasonality – 2: Mid Season Infrequent
•Phenomena – 3: Visibility <1/4 mile in
heavy snow
•Post Storm – 3 Windy and Temperatures
<32F
•Total = 11 - High
PA, Southern NY, NJ, MD
•Timing – 3: Covered multiple rush hours
•Seasonality – 2: Mid Season Infrequent
•Phenomena – 2: Moderate/heavy sleet,
wet snow or mix
•Phenomena – 3: Freezing precipitation,
black ice
•Post Storm – 2: Windy or Temperatures
<32F
•Total = 8 or 9 - High
•
October 2008 storm
Despite advances in data analysis,
assimilation and visualization,
some important user groups are not
benefiting
–
–
–
–
–
–
–
•
Very elevation dependent snowfall
>12” snow Catskills, Adirondacks
>6” snow Helderbergs, Schoharie
Valley, parts of Green Mountains
Trace of snow Capital District, Lake
George, Saratoga Regions,
Berkshires
I-84 shut down in NY/PA due to
mixed precipitation
>100,000 without power
Many trees/wires down
How can meteorologists help users
to reduce the societal impacts of
major winter storms like the
Valentine’s Day 2007 and October
2008 Storms?
–
–
–
Need to communicate forecast
information in a manner understood
by the most user groups
Need to educate users on how to
best use current forecast products
and services
Need to coordinate with users to
best tailor current and future
products for their needs
Catskills, Adirondacks and Schoharie
Valley
•Timing – 3: Covered multiple rush hours
(if rush hour exists in the mountains!)
•Seasonality – 3: First storm of season
•Phenomena – 2: Moderate/Heavy wet
snow or mix
•Post Storm – 2: Windy or Temperatures
<32F
•Total = 10 - High
Albany, valley locations
•Timing – 1: overnight
•Seasonality – No factor since no
accumulation, but could be 3 if considered
1st storm of season
•Phenomena – 1: Light intensity snow or
mix
•Post Storm – 1: Temperatures slowly
moderating above freezing
•Total = 3 – Low, but could be moderate if
considered 1st storm of season
10/08-Valleys
1
2/07
1
POD=0 FAR=1
10/08 -Mtns
2
POD=1 FAR=0
2
•Higher elevations received accumulating
snow – low impact for valley areas,
including densest population centers
•Divide into different elevations or divide
rural vs. urban– sub county?
•Limited ability to delineate sub county
areas in text products due to software
constraints
•Increasing use of graphical forecasts
needed
•All synoptic-scale storms are high impact
somewhere, so some division necessary
•Maybe consider pavement/ground
temperatures for accumulation and road
treatment factors?
Call to action statements – text or
graphics?
•Overstated threat in valley locations – any
impact due to unnecessary preparation?
•Road crews in impacted areas – best use
of resources- winter still >1 month away?
•Very wet snow – trees and power lines
down - power companies most efficient use
of resources?
•Addressing variety of vulnerability factors
such as age, health, wealth, gender?
•Graphics of hazards on map with
census/demographic data: Assist
EMs/Specialized Users – see SVR slide 18
•Largely up to broadcast media to convey
message to most of the user community
Ensemble products as guidance but
must be calibrated
Experimental Probabilistic QPF: Forecaster produced but based on POP and QPF grids –
Once ensemble and forecaster probabilities calibrated, potential use in PQPF grid
Severe Thunderstorm and Tornado Warnings
•
•
•
•
•
•
•
•
•
•
•
•
•
Current – Polygon warnings for part of a county/counties
Successive polygon warnings for locations downstream
Verified by observation of severe weather – POD, FAR, CSI
Future trends – Probabilistic severe weather information
Polygons composed of a range of probabilities for various severe
weather types
Polygons for estimated time of arrival for various severe weather types
Forecaster generated probabilities based on perceived threat –
mesoscale/storm scale analyses, radar based or observed
Short-range microscale numerical models being developed for
thunderstorm evolution predictions on the scale of minutes
Some private sector capability to produce future radar graphics for
severe thunderstorms (not presented here)
What probability would activate EAS?
Will EAS exist in the future? – Cell phones, cable TV, NYAlert
Inspiration from probabilistic hurricane wind and surge graphics
Gridded verification – including quantifying choosing not to warn
Current Methodology
Long lead time for location A
New polygon issued with short lead time
for location B
Possible future Methodology – with assistance from
meso/micro scale models (currently in development)
Polygon with range of
probabilities updated frequently,
perhaps every 5-15 minutes
In this case – rapid updates move polygon
east little by little so locations A and B
receive similar progression of information
Example of recent event – 24 July EF2
tornado in NH/ME
1533 UTC GYX radar image –
Tornado touch down, no
reports yet, SVR issued
Probabilistic polygon/accum.
threat could have provided
tornado probabilities based on
radar data
Any non-zero probability can
provide critical information to
different user groups
Example of recent event – 24 July EF2
tornado in NH/ME
1538 UTC GYX radar image
– Tornado on ground, 1st
reports beginning to come
in, TOR issued 1546 UTC
Estimated time of arrival
compliments probabilities
on previous page
Some current cutting edge severe weather
display products – Google Earth applications
Some current cutting edge severe weather
warning and verification products
Graphical and text warnings with radar and severe reports
Demographic data for each warning
Experimental verification statistics
Gridded/Probabilistic severe warning
verification
• White – forecast area
gridded by Lat/Lon
• Gray – Watch area
• Black – Polygon Warning
• Red – Severe Weather
report
• Many new statistics can
be calculated
• Red could also be radar
detected mesocyclone or
TVS
• Quantifying choice of no
warning – if warning not
issued and no severe
weather reported
• Probabilistic warnings –
even more new statistics
possible
Other considerations
•
Social science collaborations
– Emphasis on surveys and behavioral science to evaluate perceptions and
psychology – input into probabilistic forecast guidance
• What do users know/understand?
• What are their preferred information sources?
• How can the message be most clearly conveyed, minimizing misinterpretation of
the risk?
• What influences their decisions under stress and uncertainty?
• What motivates them to prepare and respond to potentially life threatening
hazards?
• What capabilities do they have to prepare and respond? (Exposure and
vulnerability issues)
– Emphasis on economic and human impacts
• NWS Service Assessments – Super Tuesday and Picher OK
• Influencing government/insurance policy - Katrina
• Input into community hazard mitigation
– Ft. Collins floods and Katrina/Ike
– Potential local collaborative study of Schoharie County NY
•
Partnering between all sectors
– AMS Ad Hoc Committee on Communicating Uncertainty in Forecasting
– National Weather Center – Government, Private Sector and Academia all in
one complex
– Hazardous Weather Test Bed
Acknowledgments
•
•
•
•
•
•
Iowa Environmental Mesonet division of the Iowa State University Department
of Agronomy
NWS Fort Worth and the North Central Texas Council of Governments
Greg Stumpf and Travis Smith for probabilistic warning graphics
Harold Brooks for gridded severe weather graphic
Howard Altshule of Forensic Weather Consultants for pictures from 28 October
2008 storm
Eve Gruntfest and WAS*IS/SSWIM founders for promoting physical and social
science collaborations that are directing future product development
Thank you for your time and attention –
are there any questions or comments?
Organizations paving the way for increased Social and
Physical Science collaborations
Web Sites
•Http://www.sip.ucar.edu/wasis - WAS*IS web site
•Http://ewp.nssl.noaa.gov/wasis2008 - Probabilistic warning workshop
•Http://eyewall.met.psu.edu – Source of many ensemble forecast guidance products