Post on 19-Jan-2016
description
Prediction of Atlantic Tropical Cyclones with the Advanced
Hurricane WRF (AHW) Model
Prediction of Atlantic Tropical Cyclones with the Advanced
Hurricane WRF (AHW) Model
Jimy Dudhia
Wei Wang
James Done
Chris Davis
MMM Division, NCAR
Jimy Dudhia
Wei Wang
James Done
Chris Davis
MMM Division, NCAR
Advanced Hurricane WRF in 2007
Advanced Hurricane WRF in 2007
Features 3 domains (12 km, 4 km, 1.33 km) 00Z start time, 5 days (domain 3 starts at 12 h) Domains 2 and 3 move with storm using vortex
following algorithm 1d ocean mixed-layer model (described later) New surface flux formulations for hurricane
conditions (described later)
Features 3 domains (12 km, 4 km, 1.33 km) 00Z start time, 5 days (domain 3 starts at 12 h) Domains 2 and 3 move with storm using vortex
following algorithm 1d ocean mixed-layer model (described later) New surface flux formulations for hurricane
conditions (described later)
1D Ocean Mixed-Layer Model1D Ocean Mixed-Layer Model
In 2007 a 1d model based on Pollard, Rhines and Thompson (1973) was added to the hurricane forecasts
Purpose is to represent cooling of sea-surface temperature due to deep mixing of the oceanic boundary layer with stably-stratified cooler water below
This has a negative feedback on hurricanes which helps to prevent over-prediction
In 2007 a 1d model based on Pollard, Rhines and Thompson (1973) was added to the hurricane forecasts
Purpose is to represent cooling of sea-surface temperature due to deep mixing of the oceanic boundary layer with stably-stratified cooler water below
This has a negative feedback on hurricanes which helps to prevent over-prediction
1D Ocean Mixed-Layer Model1D Ocean Mixed-Layer Model
Mechanism Stress from strong surface winds generates
currents in the oceanic mixed layer (typically 30-100 m deep)
Currents lead to mixing with cooler water below when Richardson number becomes low enough
This cools the mixed layer, which changes the SST and hence surface fluxes
Mechanism Stress from strong surface winds generates
currents in the oceanic mixed layer (typically 30-100 m deep)
Currents lead to mixing with cooler water below when Richardson number becomes low enough
This cools the mixed layer, which changes the SST and hence surface fluxes
1D Ocean Mixed-Layer Model1D Ocean Mixed-Layer Model
Model Slab mixed layer
Predicts depth, vector current components, temperature Initial depth specified from obs or climatology Initial temperature is from SST Initial current is zero
hurricane induced currents are assumed to dominate over pre-existing currents
Lapse rate below mixed layer is specified from obs or climatology Inertial turning effects from Coriolis are included Thermal energy balance in mixed layer due to surface fluxes and
radiation is included (small effect)
Model Slab mixed layer
Predicts depth, vector current components, temperature Initial depth specified from obs or climatology Initial temperature is from SST Initial current is zero
hurricane induced currents are assumed to dominate over pre-existing currents
Lapse rate below mixed layer is specified from obs or climatology Inertial turning effects from Coriolis are included Thermal energy balance in mixed layer due to surface fluxes and
radiation is included (small effect)
1D Ocean Mixed-Layer Model1D Ocean Mixed-Layer Model
T profile
Mixed layer
Stable layer
current
windSurface stress
Heat fluxes
SSTMixedlayer depth
mixing
1D Ocean Mixed-Layer Model1D Ocean Mixed-Layer Model
T profile
Mixed layer
Stable layer
current
windSurface stress
Heat fluxes
SSTMixedlayer depth
mixing
Surface FluxesSurface Fluxes
Heat, moisture and momentum Heat, moisture and momentumH =ρcpu*θ* τ =ρu*u*E =ρu*q*
u* =kVr
ln(zr / z0 )−ψ m
θ* =kΔθ
ln(zr / z0h ) −ψ h
q* =kΔq
ln(zr / z0q)−ψ h
Subscript r is reference level (lowest model level, or 2 m or 10 m)z0 are the roughness lengths
Roughness LengthsRoughness Lengths
Roughness lengths are a measure of the “initial” length scale of surface eddies, and generally differ for velocity and scalars
In 2006 AHW z0h=z0q are small constants (=10-4 m for water surfaces)
z0 for momentum is a function of wind speed following tank experiments of Donelan (this replaces the Charnock relation in WRF). This represents the effect of wave heights in a simple way.
Roughness lengths are a measure of the “initial” length scale of surface eddies, and generally differ for velocity and scalars
In 2006 AHW z0h=z0q are small constants (=10-4 m for water surfaces)
z0 for momentum is a function of wind speed following tank experiments of Donelan (this replaces the Charnock relation in WRF). This represents the effect of wave heights in a simple way.
Drag CoefficientDrag Coefficient
CD10 is the 10 m drag coefficient, defined such that
CD10 is the 10 m drag coefficient, defined such that
CD10 =k
ln(z10 / z0 )⎛
⎝⎜⎞
⎠⎟
2
It is related to the roughness length by (in neutral conditions)
τ ≡ρCD10V102
Enthalpy Exchange CoefficientEnthalpy Exchange Coefficient
CE10 is the 10 m moisture exchange coefficient, defined such that
CE10 is the 10 m moisture exchange coefficient, defined such that
CE10 =k
ln(z10 / z0 )⎛
⎝⎜⎞
⎠⎟k
ln(z10 / z0q)
⎛
⎝⎜
⎞
⎠⎟
It is related to the roughness lengths (assuming neutral conditions) by
Often it is assumed that CH=CE=Ck where Ck is the enthalpy exchange coefficient. However, since 90% of the enthalpy flux is latent heat, the coefficient for sensible heat (CH) matters less than that for moisture (CE)
E ≡ρCE10V10Δq
Dean track forecastsDean track forecasts
Hurricane DeanHurricane Dean
Reached category 5 in Caribbean Minimum central pressure - 906 hPa Maximum wind - 165 mph Made landfall as category 5 in Belize and
Mexican Yucatan Redeveloped over Gulf Second landfall in Mexico as category 2
Reached category 5 in Caribbean Minimum central pressure - 906 hPa Maximum wind - 165 mph Made landfall as category 5 in Belize and
Mexican Yucatan Redeveloped over Gulf Second landfall in Mexico as category 2
Hurricane Dean (2007)Hurricane Dean (2007)
Note that forecasts underestimate maximum windspeed
Hurricane Dean (2007)Hurricane Dean (2007)
Forecasts also underestimate pressure drop
CD and CkCD and Ck
From the works of Emanuel (1986), Braun and Tao (2001) and others the ratio of Ck to CD is an important factor in hurricane intensity
Observations give some idea of how these coefficients vary with wind speed but generally stop before hurricane intensity
From the works of Emanuel (1986), Braun and Tao (2001) and others the ratio of Ck to CD is an important factor in hurricane intensity
Observations give some idea of how these coefficients vary with wind speed but generally stop before hurricane intensity
Black et al. (2006)Black et al. (2006)
27th AMS Hurricane conference
Modification to Ck in AHWModification to Ck in AHW
Commonly z0q is taken as a constant for all wind speeds
However for winds greater than 25 m/s there is justification for increasing this to allow for sea-spray effects that may enhance the eddy length scales
We modify z0q in AHW to increase at wind speeds > ~25 m/s
This impacts Ck as shown next
Commonly z0q is taken as a constant for all wind speeds
However for winds greater than 25 m/s there is justification for increasing this to allow for sea-spray effects that may enhance the eddy length scales
We modify z0q in AHW to increase at wind speeds > ~25 m/s
This impacts Ck as shown next
Modification to Ck in AHWModification to Ck in AHW
Cd - red, Ck - blueNew - dashedOld - solid
Cd - red, Ck - blueNew - dashedOld - solid
Dean track forecastsDean track forecasts
QuickTime™ and aBMP decompressor
are needed to see this picture.
Impact on Dean forecastsImpact on Dean forecasts
General improvement in both wind and pressure intensity measures
This is consistent with enhanced ratio of Ck to CD as expected from previous theoretical and modeling studies
This new formulation was used for the 2007 season verifications
Impact of changing Ck in the case of Dean was greater than impact from increasing ocean mixed-layer depth
General improvement in both wind and pressure intensity measures
This is consistent with enhanced ratio of Ck to CD as expected from previous theoretical and modeling studies
This new formulation was used for the 2007 season verifications
Impact of changing Ck in the case of Dean was greater than impact from increasing ocean mixed-layer depth
Hurricane FelixHurricane Felix
Reached category 5 in Caribbean Minimum central pressure - 929 hPa Maximum wind - 165 mph Made landfall possibly as category 5 in
Nicaragua and Honduras
Reached category 5 in Caribbean Minimum central pressure - 929 hPa Maximum wind - 165 mph Made landfall possibly as category 5 in
Nicaragua and Honduras
Felix track forecastsFelix track forecasts
Tropical Storm KarenTropical Storm Karen
Did not reach category 1 Minimum central pressure - 990 hPa Maximum wind - 70 mph An example of AHW over-intensifying a
weaker storm
Did not reach category 1 Minimum central pressure - 990 hPa Maximum wind - 70 mph An example of AHW over-intensifying a
weaker storm
2007 Season Verification2007 Season Verification
From Weather Underground
2007 Atlantic Season2007 Atlantic Season
15 named storms including 5 hurricanes (Dean, Felix, Humberto, Lorenzo, and
Noel) All 5 were land-falling Only Dean and Felix exceeded category 1 Dean, Felix, Gabrielle, Humberto, Ingrid, Karen,
Lorenzo, Melissa simulations were included in the verification against other models (Mark DeMaria)
15 named storms including 5 hurricanes (Dean, Felix, Humberto, Lorenzo, and
Noel) All 5 were land-falling Only Dean and Felix exceeded category 1 Dean, Felix, Gabrielle, Humberto, Ingrid, Karen,
Lorenzo, Melissa simulations were included in the verification against other models (Mark DeMaria)
1.33 km WRF-2007 Intensity Forecast Error
0
10
20
30
40
50
60
70
80
0 12 24 36 48 60 72 84 96 108 120
Forecast Time (h)
Error (knots)
SHF5OFCLDSHPLGEMNCHGGFDLHWRFAHW
22
2017
1611
8
7
Model Intensity ComparisonModel Intensity Comparison
AHW seems to improve relative to other models at longer ranges (similar result was found in previous seasons)
Resolution must be a major factor in this
AHW seems to improve relative to other models at longer ranges (similar result was found in previous seasons)
Resolution must be a major factor in this
1.33 km WRF-2007 Track Forecast Error
0
100
200
300
400
500
600
700
0 12 24 36 48 60 72 84 96 108 120
Forecast Time (h)
Error (nm)
CLP5OFCLAVNONGPSUKMGFDLHWRFAHW
5
2017
16
11
8
22
Model Track ComparisonModel Track Comparison
AHW produces better track than GFDL, and worse than AVN, comparable with HWRF. These models all use AVN large scale.
UKMO may have a better large scale leading to better track
AHW produces better track than GFDL, and worse than AVN, comparable with HWRF. These models all use AVN large scale.
UKMO may have a better large scale leading to better track
SummarySummary
Hurricane forecasts are sensitive to surface flux treatment
A change that may represent sea-spray effects has a positive impact on intensity for strong hurricanes
Intensity verification shows the benefits of AHW at high resolution compared to other models
Accuracy of track may be limited by global model boundary conditions
Weaker storms may still be over-intensified
Hurricane forecasts are sensitive to surface flux treatment
A change that may represent sea-spray effects has a positive impact on intensity for strong hurricanes
Intensity verification shows the benefits of AHW at high resolution compared to other models
Accuracy of track may be limited by global model boundary conditions
Weaker storms may still be over-intensified