Post on 29-Mar-2015
Probabilistic Guidance for Hurricane Storm Probabilistic Guidance for Hurricane Storm Surge (P-surge)Surge (P-surge)
Arthur Taylor and Bob GlahnArthur Taylor and Bob Glahn
Meteorological Development Laboratory, National Weather ServiceMeteorological Development Laboratory, National Weather Service
January 22, 2008January 22, 2008
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Hurricane Storm Surge DamageHurricane Storm Surge Damage
• Galveston 1900 – 6,000 to 12,000 deathsGalveston 1900 – 6,000 to 12,000 deaths• Okeechobee 1928 – more than 2,500 deathsOkeechobee 1928 – more than 2,500 deaths• Florida Keys, Labor Day 1935 – 408 deathsFlorida Keys, Labor Day 1935 – 408 deaths• New England 1938 – 600 deathsNew England 1938 – 600 deaths• Audrey 1957 – 390 deathsAudrey 1957 – 390 deaths• Camille 1969 – 256 deathsCamille 1969 – 256 deaths• Hugo 1989 – 50 deathsHugo 1989 – 50 deaths• Opal 1995 – 9 deathsOpal 1995 – 9 deaths• Katrina 2005 – more than 1,800 deathsKatrina 2005 – more than 1,800 deaths Aerial Photo overlay of
Katrina 2005 storm surge over Hancock County, Mississippi
““The greatest potential for loss of life The greatest potential for loss of life related to a hurricane is from the storm related to a hurricane is from the storm surge.”surge.”
Richelieu Apartments - Before Camille 1969
Richelieu Apartments - After Camille 1969
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Storm Surge ForecastingStorm Surge Forecasting
The Sea, Lake, and Overland Surges from Hurricanes The Sea, Lake, and Overland Surges from Hurricanes (SLOSH) model is the National Weather Service’s (NWS) (SLOSH) model is the National Weather Service’s (NWS) operational hurricane storm surge model.operational hurricane storm surge model.• The NWS uses composites of its results to predict potential storm surge The NWS uses composites of its results to predict potential storm surge
flooding for evacuation planningflooding for evacuation planning
• The National Hurricane Center (NHC) begins operational SLOSH runs The National Hurricane Center (NHC) begins operational SLOSH runs 24 hours before forecast hurricane landfall24 hours before forecast hurricane landfall
The operational runs are based on a single NHC forecast The operational runs are based on a single NHC forecast track and its associated parameters.track and its associated parameters.• When provided accurate input, SLOSH results are within 20% of high When provided accurate input, SLOSH results are within 20% of high
water marks.water marks.
• Track and intensity prediction errors cause large errors in SLOSH Track and intensity prediction errors cause large errors in SLOSH forecasts and can overwhelm the SLOSH results.forecasts and can overwhelm the SLOSH results.
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Hurricane Ivan: A Case StudyHurricane Ivan: A Case Study
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Probabilistic Storm Surge Probabilistic Storm Surge MethodologyMethodology
Use an ensemble of SLOSH runs to create Use an ensemble of SLOSH runs to create probabilistic storm surge (P-surge) probabilistic storm surge (P-surge) • Intended to be used operationally so it is based on NHC’s Intended to be used operationally so it is based on NHC’s
official advisory.official advisory.• P-surge’s ensemble perturbations are determined by P-surge’s ensemble perturbations are determined by
statistics of past performance of the advisoriesstatistics of past performance of the advisories
Hurricane forecast errors which impact storm surge:Hurricane forecast errors which impact storm surge:• Cross track errors (impacts Location)Cross track errors (impacts Location)• Along track errors (impacts Forward Speed)Along track errors (impacts Forward Speed)• Intensity errors (impacts Pressure)Intensity errors (impacts Pressure)• Size of the storm errors.Size of the storm errors.
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Katrina Advisory 23Katrina Advisory 23
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Varying Katrina’s TracksVarying Katrina’s Tracks
• Include 90% of Include 90% of possible cross track possible cross track error (roughly 3 error (roughly 3 times the size of times the size of the cone of error).the cone of error).
• Spacing based on Spacing based on size of the stormsize of the storm
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Varying the Other ParametersVarying the Other Parameters
Size: Small (30%), Medium (40%), Large (30%)Size: Small (30%), Medium (40%), Large (30%)
Forward Speed: Fast (30%), Medium (40%), Slow (30%)Forward Speed: Fast (30%), Medium (40%), Slow (30%)
Intensity: Strong (30%), Medium (40%), Weak (30%)Intensity: Strong (30%), Medium (40%), Weak (30%)
The weight of a run is: cross track weight * along track weight * intensity The weight of a run is: cross track weight * along track weight * intensity weight * size weightweight * size weight
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Is P-surge Statistically Reliable?Is P-surge Statistically Reliable?
If we forecast a 20% chance of storm surge exceeding If we forecast a 20% chance of storm surge exceeding 5 feet numerous times, then on 20% of those times 5 feet numerous times, then on 20% of those times storm surge should exceed 5 feet.storm surge should exceed 5 feet.• Create a reliability diagram comparing the ratio of Create a reliability diagram comparing the ratio of
occurrence with forecast probability.occurrence with forecast probability.
Problem: Insufficient observationsProblem: Insufficient observations• Number of hurricanes making landfall is relatively small. Number of hurricanes making landfall is relatively small.
• Observations are made where there has been surge. Observations are made where there has been surge.
340 observations for storms between 1998-2005340 observations for storms between 1998-2005
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
SLOSH HindcastSLOSH Hindcast
Used SLOSH hindcast runs for “observations”.Used SLOSH hindcast runs for “observations”.• NHC used best historical information for inputNHC used best historical information for input
• Given accurate input, model results are within 20% of high Given accurate input, model results are within 20% of high water marks.water marks.
Advantage: Advantage: • Uniform “observations” everywhere, even where there is Uniform “observations” everywhere, even where there is
little or no surge.little or no surge.
Disadvantage: Disadvantage: • Same surge model used in analysis as in P-surge.Same surge model used in analysis as in P-surge.
Reliability Diagrams for Forecasts > 5 feet
48hr
2158912hr
2108517877
26042
25268
2881043648
59354
90598
176494
286559
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 10 20 30 40 50 60 70 80 90 100
Probability forecast (%) > 5 feet
Rat
io o
f O
ccu
rren
ce
5010
100
24hr
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.915064
14340
39439
4226564931
57568
71179
125725
246406
380314
0
1
0 10 20 30 40 50 60 70 80 90
Probability forecast (%) > 5 feet
Rat
io o
f O
ccu
rren
ce
1205
9107
7488
20908
3281875688
111020
212597
448559
5453960
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 10 20 30 40 50 60 70 80 90 100
Probability forecast (%) > 5 feet
Rat
io o
f O
ccu
rren
ce
1997
36hr
229
4195
11217
27574
45712
69783
90632
182758
311550
476672
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 10 20 30 40 50 60 70 80 90 100
Probability forecast (%) > 5 feet
Rat
io o
f O
ccu
rren
ce
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Probability of Probability of >> X feet of Storm X feet of Storm SurgeSurge
To calculate the probability of exceeding X feet:To calculate the probability of exceeding X feet:• For each cell, add the associated weights of the hypothetical For each cell, add the associated weights of the hypothetical
storms whose maximum surge values are greater than X storms whose maximum surge values are greater than X feet.feet.
Example: Example: • Five hypothetical storms have weights of 0.1, 0.2, 0.4, 0.2, Five hypothetical storms have weights of 0.1, 0.2, 0.4, 0.2,
and 0.1and 0.1
• The first two exceeded X feet in a given cell.The first two exceeded X feet in a given cell.
• The probability of exceeding X feet in that cell is: 30% (0.1 The probability of exceeding X feet in that cell is: 30% (0.1 + 0.2 = 30%)+ 0.2 = 30%)
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Probability of Probability of >> 5 feet of Storm 5 feet of Storm Surge for Katrina Adv 23Surge for Katrina Adv 23
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Height Exceeded by X percent of the Height Exceeded by X percent of the Ensemble of StormsEnsemble of Storms
To calculate the height exceeded by X percent of the To calculate the height exceeded by X percent of the ensemble runs:ensemble runs:• For each cell, find the surge value where the weights of the For each cell, find the surge value where the weights of the
surge values which are higher add up to a value surge values which are higher add up to a value << X. X.
Example: Example: • Five hypothetical storms have maximum surge values of 6, Five hypothetical storms have maximum surge values of 6,
5, 4, 3, 2 feet and respective weights of 0.2, 0.4, 0.1, 0.1, 0.2. 5, 4, 3, 2 feet and respective weights of 0.2, 0.4, 0.1, 0.1, 0.2.
• The height exceeded by 60% of the ensemble is 4 feet, since The height exceeded by 60% of the ensemble is 4 feet, since the 6 foot value represents the top 20% of the storms, and the 6 foot value represents the top 20% of the storms, and the 5 foot value represents the next 40%.the 5 foot value represents the next 40%.
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Height Exceeded by 10% of the Height Exceeded by 10% of the Ensemble for Katrina Adv 23Ensemble for Katrina Adv 23
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
http://www.weather.gov/mdl/psurgehttp://www.weather.gov/mdl/psurge
When is it When is it available?available?
• Beginning when Beginning when the NHC issues a the NHC issues a hurricane watch hurricane watch or warning for the or warning for the continental UScontinental US
• As close to the As close to the advisory release advisory release time as possibletime as possible
Probabilistic Storm Surge 2008Probabilistic Storm Surge 2008
Current DevelopmentCurrent Development
• We were “experimental” in 2007We were “experimental” in 2007
• The model is running in NCEP’s job stream.The model is running in NCEP’s job stream.
• The data are flowing to the National Digital Guidance The data are flowing to the National Digital Guidance Database (NDGD)Database (NDGD)
• The data will soon be available to NWS forecast offices.The data will soon be available to NWS forecast offices.
• A decision will be made soon on becoming “operational” in A decision will be made soon on becoming “operational” in 2008.2008.
• We continue to develop training material.We continue to develop training material.
• We continue to update the error statistics.We continue to update the error statistics.