bin liu, ncsu/meas/cfdl colleagues:
DESCRIPTION
Improving Hurricane Forecasting through Air-Sea-Wave Coupling and Scale-Selective Data Assimilation. Bin Liu, NCSU/MEAS/CFDL Colleagues: Drs. Lian Xie, Huiqing Liu, Shiqiu Peng, and other current and previous CFDL group members. NOAA GOES-13 satellite Imagery of Hurricane Irene. - PowerPoint PPT PresentationTRANSCRIPT
NOAA GOES-13 satellite Imagery of Hurricane Irene
Improving Hurricane Forecasting
through Air-Sea-Wave Coupling
and Scale-Selective Data
AssimilationBin Liu, NCSU/MEAS/CFDL
Colleagues:Drs. Lian Xie, Huiqing Liu, Shiqiu Peng, and other current and
previous CFDL group members
TC forecast error trends for the Atlantic Basin
From http://www.nhc.noaa.govThe specific goals of the HFIP are to reduce the average errors of hurricane track and intensity forecasts by 20% within five years (2014) and 50% within 10 years (2019) with a forecast period out to seven days. (From HFIP’s Strategic Plan)
Outline
• Air-Sea-Wave Interaction– A Coupled Atmosphere-Wave-Ocean Modeling System
(CAWOMS)• Limited-Area Model Nesting/Downscaling
– A Scale-Selective Data Assimilation (SSDA) technique
Atmospheric Boundary Layer
Oceanic Boundary Layer
From CBLAST Project
The CAWOMS
Schematic of the coupled atmosphere-wave-ocean modeling system
The Model Coupling Toolkit
• The coupling between the model components is carried out by using the Model Coupling Toolkit (MCT, Larson et al., 2005; Jacob et al., 2005)
MCT
Model Comp 1t=t0
Init MCT…
t=cpltime…
t=tendFinalize MCT
Master begin
Master end
Model Comp 2t=t0
Init MCT…
t=cpltime…
t=tendFinalize MCT
Atmosphere-ocean coupling
• Atmospheric model drives ocean model– WRF drives POM through
atmospheric forcing
• Oceanic feedback effects– POM provides sea surface
temperature to WRF to estimate air-sea heat fluxes
– POM provides sea surface currents to determine the relative wind speed for estimation of sea surface wind stress
Wave-ocean coupling
• Wave-ocean interaction– Doppler shift effect of background
current on waves– Water level variation changing water
depth – Modifying upper-ocean currents
through radiation stress– Wave-enhanced bottom stress due to
wave orbital velocity– Wave-enhanced upper-ocean mixing
due to wave breaking– Both wave state and surface ocean
conditions impact air-sea momentum and heat fluxes
Atmosphere-wave coupling
• Atmospheric model drives wave model– WRF drives SWAN through surface wind– WRF provide surface variables for
estimation of sea spray fluxes
• Wave-related feedback effects– Wave state and sea spray affected sea
surface roughness – Wave state affected sea spray heat
fluxes– Dissipative heating
Wind stress
Drag coefficient Sea surface roughness
0
* lnz
zuU zn
)/exp( 2/10 DznCzz
2)(z
Dz U
uC
2
0
2 )ln(
z
zCDzn
2 2* Du w u C U
Air-sea momentum flux
Based on the dependence of sea surface roughness on wind speed, Charnock (1955) proposed the famous Charnock relation
E.g., Wu (1980):
A constant Charnock parameter corresponds to the drag coefficient increasing linearly with wind speed.
Sea surface roughness
2*0 /ugz
0185.0
Cd-U10 relationships
Wave-age-dependent sea surface roughness
• The SCOR workgroup 101 (Jones and Toba 2001) presented a relation between the Charnock parameter and wave age (SCOR relation):
• According to this relation, the non-dimensional sea surface roughness first increases and then decreases with the increasing wave age.
* * *
*
0.03 exp 0.14 , ~ 0.35 35
0.008, 35
The SCOR relation and its comparison with various field and laboratory observations (from Toba et al., 2006)
Sea surface under hurricane winds
Photo by Mike Black, NOAA/AOML/HRD
Wave state and sea spray affected sea surface roughness
Instead of using a constant Charnock parameter, we combined the SCOR relationship (Jones and Toba 2001) with the resistance law of Makin (2005), which took into account of sea sprays impact on air-sea momentum flux, and obtained a sea surface roughness applicable to both low-to-moderate and high wind conditions:
1/3/ 2 1 1/* * * *
1/1 1/*
(0.085 ) 0.03 exp 0.14 , 35
17.60 0.008 , 35
* */pc u *min 1, cra u -10.64 m scra
Wave state and sea spray affected sea surface wind stress
The wave state and sea spray affected Cd-U10 relationship together with the field and laboratory observations
Wave state and sea spray affected sea surface scalar roughness
• As for the sea surface heat and moisture fluxes, we use the parameterization of sea surface scalar roughnesses in COARE algorithm V3.1 (Fairall et al., 2003) to estimate the direct air-sea heat fluxes.
4 5 0.6*min 1.1 10 , 5.5 10 ReT qz z
* 0 *Re /z u The Reynolds number of sea surface aerodynamic roughness
Sea sprays and air-sea heat fluxes
Sea spray droplets
From Andreas (1995)
Droplet Evaporation Layer (DEL)
From Andreas and DeCosmo (2002)
Wave state affected sea spray heat flux
• By using the wave state dependent Sea Spray Generation Function (SSGF) and Andreas (1992)’s algorithm to estimate nominal sea spray heat flux, and considering the feedback effects
• One can now estimate the sea spray sensible and latent heat fluxes that include wave state effect. and are determined following Bao et al. (2000), while is taken as 1 (Andreas, 1992).
,
,
S T S S L
L T L L
H H Q Q
H H Q
,
,
S sp S L
L sp L
Q Q Q
Q Q
Sea state affected air-sea heat fluxes
The given atmospheric and sea surface environment: sea surface pressure: 1000 hPa; air temperature: 25 °C; air relative humidity: 80%;neutral stable atmospheric layer; sea surface temperature: 27 °C; and ocean salinity: 34 psu.
CAWOMS - Idealized TC
• A bogus vortex with a maximum wind speed of 30 m s-1 at a radius of 70 km is implanted in the ambient atmosphere with uniform easterly winds of 5 m s-1.
• The ambient temperature and humidity (salinity) profiles are derived from the monthly averaged vertical profiles for September at the location of (20N, 145E):– NCEP-DOE AMIP-II Reanalysis– One degree WOA05
• The initial SST equals 29 Celsius degree, with a mixed layer depth (MLD) of 40 m.
Typical tropical atmosphere
Typical tropical ocean
u = -5 m/s
Vortex: max wind 30 m/s
Experiment design for an idealized TC
Summary of the experiments
Expts.
Couplings considered
Atmosphere-wave Atmosphere-ocean Wave-ocean
CTRL No No No
CPLAW Yes No No
CPLAO No Yes No
CPLAWO Yes Yes Yes
The simulated TC tracks
45-h results of the control run
Vertical cross section of potential temperature and horizontal wind speed.
CTRL
CPLAO
CPLAW
CPLAWO
CTRL
CPLAO
CPLAW
CPLAWO
CTRL
CPLAO
CPLAW
CPLAWO
Wind and SLP SWH SSC
Effects of atmosphere-wave-ocean coupling
CTRL
CPLAO
CPLAW
CPLAWO
Left: SSTMiddle: MLDRight: HHC
CTRL
CPLAW
CPLAO
CPLAWO
Left: Total upward sensible heat fluxRight: Total upward latent heat flux
CPLAW CPLAWO
CPLAWOCPLAW
HSS
HLL
HSD
HLD
HEE
Effects of atmosphere-wave-ocean coupling
Time series of the simulated (a) minimum SLP, (b) max 10-m wind, (c) maximum SWH, and (d) minimum SST for each experiment.
CTRL: Blue
CPLAW: Red
Expts. Min SLP Max 10-m wind Max SWH Min SST Max SSC Max MLD
CTRL 959.29 41.11 17.83 25.28 0.89 132.48
CPLAW 950.98 49.10 19.07 24.89 0.95 133.70
CPLAO 966.35 40.13 15.98 26.17 0.81 98.83
CPLAWO 962.36 42.05 16.09 25.89 0.89 107.90
CPLAO: Green
CPLAWO: Black
Sensitivity to MLD
*80 => initial MLD of 80 m*120 => initial MLD of 120 m
Summary of AWO interaction and coupling
• A CAWOMS consists of WRF, SWAN and POM, based on atmosphere-wave, atmosphere-ocean, and wave-current interaction processes.
• The AW coupling has overall positive contribution, which strengthens the TC system.
• The AO coupling has overall negative contribution due to the negative feedback of SST cooling, which weakens the TC system.
• The overall effects of AWO coupling on TC intensity depend on the balance between wave-related overall positive feedback and oceanic overall negative feedback.
Outline
• Air-Sea-Wave Interaction– A Coupled Atmosphere-Wave-Ocean Modeling System
(CAWOMS)• Limited-Area Model Nesting/Downscaling
– A Scale-Selective Data Assimilation (SSDA) technique
SSDA—A nesting/downscaling technique
GCM LAM Sub-grid scale
SSDA
L k1000 km 100 km
100 km 12 kmResolution
10000 km 10 km
Traditional sponge zone method and the SSDA approach
Domain boundary
Relaxation zone
Domain interior
The SSDA system and procedure
LAM-WRFt
Components: LAM (WRF), 3DVAR (WRFDA), Low- and band-pass filter
Application to regional climate simulation of North Atlantic (Jun-Oct
2005)
The large-scale wind field at 200 hPa for (a) GFS analysis; (b) control run; and (c) SSDA5 at 00 UTC 24 September 2005 (right after the data assimilation cycle).
CTRLGFS
SSDA
Application to TC track hindcastingHurricane Katrina (2005)
Simulated storm tracks of Hurricane Katrina for experiments (blue ‘o’) CTRL, (green ‘+’) FDDA, and (red ‘x’) SSDA, together with (black ‘s’) the best track
CTRL: Blue
SSDA: Red
BEST: Black
Application to TC track hindcastingHurricane Katrina (2005)
Simulated storm tracks: (‘o’) CTRL1, (‘*’) SSDA1, (‘+’) CTRL2, (‘x’) SSDA2, together with (‘s’) the best track
Model domains
CTRL1: Blue ‘o’SSDA1: Blue ‘*’CTRL2: Green ‘+’SSDA2: Green ‘x’BEST: Black
Application to TC forecastingHurricane Felix (2007)
Mean track (left) and intensity (right) forecast errors for the CTRL and SSDA runs, the GFS global forecasts, the CLIPPER5 or SHF5 model, and the official NHC forecasts, at different forecast periods.
Track Intensity
Application to Hurricane Irene (2011)’s
track simulation
Application to Hurricane Irene (2011)’s
intensity simulation
The Hybrid SSDA System
LAM-WRF
Components: LAM (WRF), 3DVAR (WRFDA), Low- and band-pass filter
Hybrid SSDADA for Hurricane Irene
OBS Surface OBS Upper Air
Hybrid SSDADAHRD for Hurricane Irene
H*Wind TDR wind
Hybrid SSDA for Hurricane Irene
Blue: CTRLGreen: SSDARed: SSDADACyan: SSDADAHRDPurple: BEST
Hybrid SSDA for Hurricane Irene
Valid at 12Z25AUG
CTRL SSDA
SSDADA SSDADAHRD
Summary of SSDA• The SSDA approach drives the regional model from the model
domain interior in addition to through the conventional sponge zone boundary conditions.
• The SSDA approach can be applied in various situations, including regional climate downscaling and tropical cyclone forecasting. The SSDA approach can effectively improve the TC forecasting by assimilating large-scale information from global model results, especially for the track forecasting.
• The Hybrid SSDA approach has an additional function (merit) to assimilate all kinds of available observations, leading to improvements for both TC intensity and track forecasting.
A Hybrid SSDA-CAWOMS
CAWOMS SSDADA
Hybrid SSDA
TC Intensity Forecasting
TC Track Forecasting
TC Track and Intensity Forecasting
Hybrid SSDA-CAWOMS
Thank you!