Sensitivity of the HWRF model prediction for Hurricane Ophelia (2005) to the choice

of the cloud and precipitation scheme

Yuqing Wang and Qingqing LiInternational Pacific Research center

University of Hawaii at Manoa, Honolulu, HI 96822

The 63rd Interdepartmental Hurricane ConferenceSt Petersburg, Florida, March 2-5, 2009

Acknowledgments: Naomi Surgi, Steve Lord, HWRF team as NCEP/EMC

• The storm size is generally too large (increase with time);

• Large storms are too strong while small storms are too weak;

• Storms are too energetic and hard to dissipate;• It performs best for storms in weak shear environment; • Problems in mid-upper level structure for storms in

vertical shear environment.

Some systematic biases of HWRF

Liu et al. 2008, originally from Biju Thomas

Objectives

• To identify the model physics that are critical to the structure and intensity changes in the HWRF model;

• To improve the representation of those model physics to achieve improved prediction of hurricane structure and intensity changes by HWRF model.

Working Hypothesis

• 3D distribution of diabatic heating due to phase changes is the key to both the structure and intensity of hurricanes;

• The vertical heating distribution in in eyewall determined the rate of intensity change, while horizontal heating distribution determines the storm size change;

• Realistic representation of 3D diabatic heating due to phase changes is the fundamental to any model to achieve improved prediction for hurricane structure and intensity!

Hurricane Structure and Intensity Change

Three-dimensional distribution ofInternal atmospheric heating

Grid-scale cloud microphysics

Subgrid-scale Cumulus convection

Vertical motionsPDF in grid scale and

updrafts in plumes

Nonlinear Feedbacks

Hydrometeors in updraft plumes

Initiation of clouds(Liquid/ice)

Nonlinear Feedbacks

How sensitive the simulated hurricane size is to heating in the outer spiral rainbands of the hurricane in the nonhydrostatic hurricane model TCM4

Cloud top brightness temperatures (in Celsius) from satellite observation for Hurricane Ophelia (2005) at 18 UTC 12 September 2005.

Model WRF-NMM

Experiment name NKF KF NBM BM

Convective parameterization

D1: Kain-Fritsch schemeD2: none

D1: Kain-Fritsch schemeD2: Kain-Fritsch scheme

D1: Betts-Miller-Janjic schemeD2: none

D1: Betts-Miller-Janjic schemeD2: Betts-Miller-Janjic scheme

Horizontal resolution

Mesh 1： 0.25° × 0.25° (108 × 180 × 38)Mesh 2： 0.0833° × 0.0833° (172 × 226 × 38)

PBL scheme Mellor-Yamada-Janjic TKE scheme

Precipitation scheme

Ferrier microphysics scheme

Radiation Shortwave and longwave radiation schemes of GFDL

Land surface LOAH land surface model

Lateral boundary and initial data

FNL Data

Initial time 00 UTC 09 Sept. 2005Integration 96 hours

Numerical model settings and experimental design

The model domain used in all experiments. The outer domain D1 is 0.25o resolution and the inner domain D2 is 0.08333o resolution

(a) (c)

(b) (d)

Observed (black) and simulated (red) tracks of Hurricane Ophelia (2005) in experiments (a) NKF, (b) KF, (c) NBM, and (d) BM, respectively, with marks at 6-h intervals.

Time (h)

Min

imum

sea

leve

lpre

ssur

e(h

Pa)

Max

imum

sust

aine

dw

ind

(m/s

)

0 12 24 36 48 60 72 84 96955

960

965

970

975

980

985

990

995

1000

1005

1010

15

20

25

30

35

40

45

50

Observed minimum sea level pressureObserved maximum sustained windModeled minimum sea level pressureModeled maximum sustained wind

NKFa

Time (h)

Min

imum

sea

leve

lpre

ssur

e(h

Pa)

Max

imum

sust

aine

dw

ind

(m/s

)

0 12 24 36 48 60 72 84 96955

960

965

970

975

980

985

990

995

1000

1005

1010

15

20

25

30

35

40

45

50

Observed minimum sea level pressureObserved maximum sustained windModeled minimum sea level pressureModeled maximum sustained wind

NBMc

Hour (h)

Min

imum

sea

leve

lpre

ssur

e(h

Pa)

Max

imum

sust

aine

dw

ind

(m/s

)

0 12 24 36 48 60 72 84 96955

960

965

970

975

980

985

990

995

1000

1005

1010

15

20

25

30

35

40

45

50

Observed minimum sea level pressureObserved maximum sustained windModeled minimum sea level pressureModeled maximum sustained wind

KFb

Time (h)

Min

imum

sea

leve

lpre

ssur

e(h

Pa)

Max

imum

sust

aine

dw

ind

(m/s

)

0 12 24 36 48 60 72 84 96955

960

965

970

975

980

985

990

995

1000

1005

1010

15

20

25

30

35

40

45

50

Observed minimum sea level pressureObserved maximum sustained windModeled minimum sea level pressureModeled maximum sustained wind

BMd

96-h evolution in the maximum 10-m wind speed (dashed in m s-1) and the central sea level pressure (solid in hPa) of Hurricane Ophelia (2005) from observation (red) and simulations (blue); (a) NKF, (b) KF, (c) NBM, and (d) BM.

Cloud top brightness temperatures (Celsius) simulated in experiments (a) NKF, (b) KF, (c) NBM, and (d) BM at 18 UTC 12 September 2005.

NKF NBM

BMKF

CAPE simulated in experiments (a) NKF, (b) KF, (c) NBM, and (d) BM at 18 UTC 12 September 2005.

Proposed Work

• Fast and slow cloud microphysics processes should not be equally weighted and the sedimentation of cloud ice should not be neglected. – Both would make the cloud microphysics scheme less time-

step/resolution dependent and producing more realistic 3D distribution of diabatic heating due to phase changes.

• The growth and nucleation of liquid and ice clouds depends strongly on grid-scale vertical motion and subgrid-scale turbulence, critical to horizontal extent of diabatic heating. – To take into account the subgrid-scale super-saturation in both

stratiform and convective parameterization schemes is critical to realistic simulation of cloud structure and heating due to phase changes.