TO BLEND OR NOT TO BLEND In the Pursuit of Finding an Operational Way to Give hurricane characteristics to the CMC Forecast Wind Field

Size: px

Start display at page:

Download "TO BLEND OR NOT TO BLEND In the Pursuit of Finding an Operational Way to Give hurricane characteristics to the CMC Forecast Wind Field"

Bernadette Hunter
5 years ago
Views:

1 TO BLEND O NOT TO BLEND In the Pursuit of Finding an Operational Way to Give hurricane characteristics to the CMC Forecast Wind Field Serge Desjardins oop Lalbeharry Allan MacAfee Hal itchie 3 Garry Pearson Environment Canada : Science and Technology Branch : Meteorological Service of Canada 3 : echerche en Prévision Numérique National Lab for Marine and Coastal Meteorology Halifax, Nova Scotia

2 Towards an operational hurricane surge/wave forecast system In the longer term, improvements in observations, data assimilation and NWP forecast systems should produce more accurate hurricane and extratropical transition (ET) forecasts. In search of practical improvements in the shorter term, we propose to blend parametric hurricane wind and pressure fields based on Canadian Hurricane Centre trajectory forecasts into the operational surface fields used as input for the ocean wave model and storm surge model. Because of the unpredictable nature of hurricanes, a human intervention tool is needed. The Canadian Hurricane Centre forecast trajectory becomes the official and final hurricane forecast (track and intensity) for various users. HUSWIM has been developed to supply wave and storm surge forecast guidance for forecasters when hurricanes or tropical cyclones affect the Canadian waters of responsibility.

3 A glimpse at the Blending Methodology WE Environment Weight Blending area f h f ζ c ζ c f h 0 f f ζ, P P, UV UV c 0 max h max Parametric fields GEM fields hurricane_ x : resolution 0. deg Model _ x : resolution 0.5 deg t _ blending: hour Model_ x : resolution 4km(Juan) 5km(Wilma)

4 CONCLUSIONS BLENDING methods are applied as an artificial way to merge together valuable information from different sources. One should not expect to simulate reality perfectly. Blending methods can vary and adopt personal characteristics. Insertion of a Hurricane wind field into the egional GEM forecast gives a more realistic wind forecast reflecting the presence of an intense and compact wind system. Consequently, improves the wave field. Overall, HUSWIM could help the forecaster by supplying products where the forecast wind field, used by a wet model, has a hurricane or tropical cyclone in it. The New Environment Weight parameter : Minimizes the impact of blending in developed synoptic system Allow blending of various tropical system (TS, Hurricane, ET) The New Blending adius parameter : Edge of the positive surface geostrophic vorticity seems to define well the blending zone.

hours -Generates F_H(dtB) from Hurricane Parametric Wind Model (HPWM) SWIM ATMO FEEDE 3 Hrl_F_A Hrly_F(dtA)

5 SWIM Surface Wind Interpolator and Modifier Hurr. Wind Generator - GETS Operational trajectory -Interpolates trajectory (dtb) Acceleration kept constant for 6 hours -Generates F_H(dtB) from Hurricane Parametric Wind Model (HPWM) SWIM ATMO FEEDE 3 Hrl_F_A Hrly_F(dtA) PASSIVE COM Coupler Server WET MODEL -> F_M(dt) BLENDE -> F_A(dtA) F_B(dtA) IF HUCN F_B(dtA) -> F_B(dtB) Gets F_H(dtB) F_M(dtB) (- α) F_B(dtB) + α F_H(dtB) ELSE F_M(dt)F_B(dtA)

6 Hurricane Parametric Wind Model (HPWM) Adjustments of modelled winds SLOSH Model (Jelesnianski et al. 99) Empirical: Curve Fitting method V V T V m V ( Storm m + ( m m + ) m Atlantic HPWM particularity ) adius of Max Winds (km) corresponds to mean boundary layer or gradient wind above the surface -Adjusted to 0-m elevation with V 0 K m V Km -> [70-85] % In our case Km 75% (Allan MacAfee s work) 0 Weak <7 Gale Storm Cat Cat Cat Storm Intensity (m s - ) - adii of maximum wind ( m ) curves, extracted along radial profiles from the storm centre at.5 o intervals, for different classes of storm intensity. (Storm data: HD gridded winds for 389 storms from ). - Vary with latitude Cat Cat-5 >69

7 Blending Zone : Geostrophic vorticity edge (ζ min *0-6 sec - ) influ : adius of Influence A D B LAPLACIAN COMPUTATION x ~00km ( ) 0.5* + + x C Color : Forecasted Geo. Vorticity (0-6 s - ) Black : Forecasted Isobars (4 hpa) Gray : Inserted Hurricane Isobars (4 hpa) inf lu ζ c ζ ζ c where AVG i where i indicates Point s min A,B,C,D

8 Blending Weight : Ω WL Confidence Indices W_edge (uv,p) WL 40 % W( uv) + 60 % W( p) Vortex max speed difference if uv > 5 knots mdl W( uv) uv hur uvmdl. else uvmdl Vortex pressure difference W ( p) p p p hur Lim anl.0 p p p < Lim p Lim.0 30hPa Β Ω Ω * 0.0 r r r r [, ] Ω(r,,WL) exp ( ln(wl) * ) Z Z max r ( 6 deg)

Blending : Track Dependency LEFT OF TACK : ELLIPTIC BLENDING ZONE _ ellp sin.0 Β Ω * 0.0 b ( θ φ) + ( θ ) b cos D s _ ellp _ ellp r r r r [, ] _ ellp D :Storm direction s _ ellp b 4 3.5 3.5 _III 3.

9 Blending : Track Dependency LEFT OF TACK : ELLIPTIC BLENDING ZONE _ ellp sin.0 Β Ω * 0.0 b ( θ φ) + ( θ ) b cos D s _ ellp _ ellp r r r r [, ] _ ellp D :Storm direction s _ ellp b _III 3.5 deg _II.5 deg _I deg Variation of b with the Storm speed cosφ b φ φ V extm ( + ) V s V φ extm max V s V max + arccos V s ( ) D s,v s.5 V V V min max V max Storm Speed (knots) Β.0 Ω * 0.0 r r r [, ] r

Blending : Environment Weight HUICANE TAJECTOIES HUICANE CASES CASE PEIOD

Ophelia_005 Wilma_005 Alex_004 Fabian_003 Juan_003 BDE_998 Jul. 9 - Aug.

4 - Oct. 8 Jul. 3 - Aug. 06 Sep. 04 - Sep. 08 Sep. 5 - Sep. 30. Aug. 3 - Sep.

10 Blending : Environment Weight HUICANE TAJECTOIES HUICANE CASES CASE PEIOD INTENSITY Chantal_007 Beryl_06 Florence_06 Isaac_006 Franklin_005 Ophelia_005 Wilma_005 Alex_004 Fabian_003 Juan_003 BDE_998 Jul. 9 - Aug. 0 Jul. 8 - Jul. 3 Sep Sep. 5 Sep.7 - Oct. 03 Jul. - Jul. 3 Sep Sep. 9 Oct. 4 - Oct. 8 Jul. 3 - Aug. 06 Sep Sep. 08 Sep. 5 - Sep. 30. Aug. 3 - Sep. 09 TP TS TS-SS-TS TS-SS-TS TS TS/SS-TS SS-SS3-TS TS-SS/-TS SS4-SS TS-SS SS-TS / TS

11 Blending : Vortex Distance Dependency FINAL ADJUSTMENT on and WE BLENDING PAAMETES D influ A C WE B Λ X WE r 0. 6 deg Β WE * Ω * 0.0 M r M r r M [ M, ]

12 Juan 003 : Sep 5/00z - 30/00z Wilma 005 : Oct 4/00z Oct 8/00z

13 Blending : Parameter Evolution

14 Blended Fields : Juan : Sep PESSUE AND WIND SIG WAVE AND SWELL VOTICITY IF INTEESTED I HAVE ANIMATIONS FO ALL CASES

15 Scatter Plots for Juan and Wilma Wilma (black) and Juan (red) SWH and Wind Speeds Scatter Plots Legend: HU Hurricane blending NOHU No hurricane blending CG Coarse grid (0.5 o res.)

16 Wind /Wave time series : Juan Wind Speed Wave Heigth Legend: HU Hurricane blending NOHU No hurricane blending CG Coarse grid (0.5 o res.)

17 Wind /Wave time series : Juan 4437 Wind Speed Wave Heigth Legend: HU Hurricane blending NOHU No hurricane blending CG Coarse grid (0.5 o res.)

18 Wind /Wave time series : Wilma Wind Speed Wave Heigth Legend: HU Hurricane blending NOHU No hurricane blending CG Coarse grid (0.5 o res.)

19 Wind /Wave time series : Wilma Wind Speed Wave Heigth Legend: HU Hurricane blending NOHU No hurricane blending CG Coarse grid (0.5 o res.)

20 FUTUE WOK Technical side : Enlarging the model domain inside HUSWIM for the laplacian computation. Toward an operational implementation for next hurricane season Studying more past cases (blended versus observations) efining the methodology Targeting more higher impact events. educing the jump in the vortex distance (caused by model flip-flop?) Tuning the environment weight function Publication of the results

BLENDING PARAMETRIC HURRICANE SURFACE FIELDS INTO CMC FORECASTS AND EVALUATING IMPACT ON THE WAVE MODEL FOR HURRICANE JUAN

BLENDING PARAMETRIC HURRICANE SURFACE FIELDS INTO CMC FORECASTS AND EVALUATING IMPACT ON THE WAVE MODEL FOR HURRICANE JUAN Serge Desjardins Roop Lalbeharry Allan MacAfee Hal Ritchie 8th International Workshop