The Local Wind Forecast and Its Uncertainty Sessions/TS23 Trop... · 2016. 2. 29. · Pablo Santos...
Transcript of The Local Wind Forecast and Its Uncertainty Sessions/TS23 Trop... · 2016. 2. 29. · Pablo Santos...
The Local Wind Forecast and Its Uncertainty
(Completing the Wind Forecast)
Florida Governor Hurricane ConferenceMay 11-16
2014
Pablo Santos
Meteorologist In Charge
NWS Miami/South Florida Forecast Office
2014 – FLGHC
TS 23
Motivation
• What is there in it for me in this presentation?
– A better appreciation of the approach used by local offices to downscale NHC’s forecast to provide you more detailed locally driven information.
– A better understanding of the limitations behind deterministic only based information and basing decisions on those alone.
– A better understanding of the wind speed probabilities available to you nationally and at the local level and:
• What they mean
• What questions you might have that THEY CAN answer
• How NWS can help you make better decisions using this data
• And potential advanced applications.
NHC Wind Forecasts
Full Radii Only Out to 36 Hrs
50/34 kts Radii Only Out to 72 Hrs
No Radii Days 96/120 Hrs
NHC Official Forecast
WFO Wind Grid Forecasts
WFO Wind Grid Forecasts
NHC/WFO Wind Forecasts
• So when making forecast of Wind locally using NHC guidance, NWS offices then have to refine that forecast to create a wind grid based on:
– Conceptual models– Climatology-based data – Local knowledge– Understanding of smaller scale processes– Diagnostic data for present conditions – Other tools such as high resolution models, when available
• Intensity changes are depicted as a steady trend in space and time.
• Wind Speeds can be artificially reduced prior to landfall as they are interpolated within 12-hour forecast points. They are worse in days 4 and 5.
• We are required to make wind grids forecasts at the hourly scale out to 48 hours, 3 hourly out to 72 hours, and 6 hourly out to 120 hours with this guidance on a 2.5 km grid.
• The limitations (which translate into errors) of the assumptions listed so far are magnified the higher the resolution of the wind grids.
• This is how the local offices create a local wind forecast from the NHC forecasts. ASSUMPTIONS AND LIMITATIONS RESULT IN GREATLY MAGNIFIED ERRORS when looking at deterministic wind forecasts alone. HURRIVAC has the same limitations.
NHC/WFO Wind Forecasts Limitations
Local Wind ForecastWhat is the Point?
• Approach currently used is science-based, but still limited with uncertainty and many sources of errors. All of these limitations also apply to Hurrevac.
• We are delivering information with greater precision, BUT IT DOES NOT IMPLY ACCURACY. Science is not where we would like it to be; service is outpacing the state of the science.
• This is in part why NHC delivers their advisory forecasts the way they do, but the need to provide more detailed local level info is pushing the envelope.
• Message From a Decision Making Perspective:
– AVOID OVER-RELYING ON DETERMINISTIC ONE SCENARIO/BEST GUEST ALONE!
• It is bad enough to rely in one scenario alone only without considering the issues raised so far.
– Be well informed about these issues and embrace a combined official forecast with probabilistic-based information.
D-Day Give me the Probabilities
1111
Wind Speed Probability Text Product(abbreviated)In addition to:
- Onset Probabilities: Text
- Cumulative Wind Speed Probability: Text/Graphic
11
Onset Period ProbabilitiesCumulative Probabilities
What Questions do They Answer?
• Cumulative (Available in NHC Graphic and PWSAT#; coarse time resolution) – What arethe chances that tropical storm or hurricane conditions will occur between hour 00 andXX out to 120 hours/5days with this event at my location?
• Onset (Available in PWSAT#; coarse time resolution) – What are the chances thattropical storm or hurricane conditions will begin during a particular time period at mylocation? And given that, what is the most likely period of onset of these conditions?
• Incremental (Not available from NHC; but from NWS offices) – What are the chancesthat tropical storm or hurricane conditions will be experienced during a particularperiod at my location? How likely is the event to happen during that period? How likelyis it to last? At what values is the event becoming more plausible (likely) than justpossible?
We consider trend from advisory to advisory
Cumulative
Incremental – 12HR
Onset
Onset
Tropical
Storm
Conditions
Onset
Hurricane
Conditions
Threshold:
45%
Threshold:
35%
Threshold:
30%
Threshold:
25%
Threshold:
22.5%
Threshold:
20%
Threshold:
17.5%Incremental
Tropical
Storm
Threshold:
25%
Threshold:
20%
Threshold:
15%
Threshold:
12.5%
Threshold:
10%
Threshold:
8%
Threshold:
7%Incremental
Hurricane
Local ScaleWind Speed Probabilities
Consider Trend From Advisory to
Advisory
Applications
• Communicate Risk.
• Provide objective measure of uncertainty that can be used to create new targeted products.
– Threat/Potential Impact Graphics• Briefings
• Can be used as decision making aid by planners (example follows).
• Communicate to the public graphically degree of preparation needed.
• Can be used by EM and Media along with NWS to help communicate the bottom line message to the public.
– Expressions of Uncertainty• Used to enhance forecast information during tropical cyclones (Expressions of Uncertainty).
• Can be used as decision making aid by planners.
• For trend analysis from advisory to advisory for proper risk assessment.
Cumulative: Note that
chances of hurricane-
force winds at Tampa
Bay and Port Charlotte
are both around 30%!
Tampa
Port Charlotte
IncrementalNotice: highest right
up Port Charlotte
Communicate RISK
Example: Forecast Applications: Expressions of UncertaintyAdvisory Time: 20040902_1500 (25-36 Hours – Third Period)
FRANCES (ZFP) FRANCES (Click Point)
Official in 2014
Example: A major hurricane approximately 36-hours from
projected landfall (Hurricane Warning issued). Deterministic
NHC track/intensity forecast used
by WFO’s to create local wind
forecasts (location/timing specific)
NHC
WFO’s
NHC 34 kt probability
NHC 64 kt probability
Probabalistic
NHC wind probabilities
also incorporated to
produce Impact Graphics
Threat/Potential Impact Graphics
Official Forecast Actual Threat/Potential Impacts
Example: Approaching major hurricane at the onset of the warning
period (~ 36 hours). Remember to consider the trend with these from
advisory to advisory also. Notice implicit use of probabilistic information.
QUESTION: When advocating the measure of protective
actions according to wind impacts, which is better?2014 - FLGHC
Threat/Potential Impact Graphics
Storm Intensity
Cat 3(96-113 Kt)
Threat?
OR
FORECAST VALID XX/XX00Z 31.4N 77.5W
MAX WIND XXX KT...GUSTS XXX KT.
Neal Batista
Decision Making Guide
Clearance Times
Forecast impact ≤33hrs?
FORECAST VALID XX/XX00Z 31.4N 77.5W
MAX WIND XXX KT...GUSTS XXX KT.
Available 2012 season
Neal Batista
Wind Probabilities
64KtIncremental wind
prob ≥25%?
Neal Batista
More on Trend –Wilma Example
Impact Graphics use
implicitly probabilistic
guidance. What about
their trend from
advisory to advisory
also?
2014
Tropical
Workshop
• Advisory 30• Advisory 30
• Advisory 31
• Advisory 32
• Advisory 35
Take Away
• Be aware of the data upon which you base your decisions (whether national or local):
– Your NWS office can help you sort through it and target your needs with their proper application.
• Be aware of the limitations of deterministic only based information.
– And making decisions solely based on deterministic or even “alternate scenarios” only.
• Be aware of all probabilistic data available to you including
– Cumulative, Onset, Incremental
– Questions they answer
– Their temporal resolution
– Be knowledgeable of their significance and trend behavior.
– Again your local NWS office can help you sort through this
• Be aware of the potential applications of using combined deterministic/probability data and their utility in communicating the message and making decisions.
– Again, your local NWS office can help you with this
Questions…Pablo Santos
Meteorologist In ChargeNWS Miami/South Florida Forecast Office