+ All Categories
Home > Documents > Investigating a perturbed physics scheme in a wave ensemble system

Investigating a perturbed physics scheme in a wave ensemble system

Date post: 03-Jan-2016
Category:
Upload: jescie-solis
View: 63 times
Download: 0 times
Share this document with a friend
Description:
Investigating a perturbed physics scheme in a wave ensemble system. Ray Bell (Line manager: Francois-Bocquet) Ocean Iced Tea 14/09/2010. Table of Contents Wave ensemble system Perturbed physics scheme Sensitivity studies – Effective wind stress - PowerPoint PPT Presentation
37
© Crown copyright Met Office Investigating a perturbed physics scheme in a wave ensemble system Ray Bell (Line manager: Francois-Bocquet) Ocean Iced Tea 14/09/2010
Transcript
Page 1: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Investigating a perturbed physics scheme in a wave ensemble system Ray Bell (Line manager: Francois-Bocquet)

Ocean Iced Tea 14/09/2010

Page 2: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Table of Contents

• Wave ensemble system

• Perturbed physics scheme

• Sensitivity studies – Effective wind stress

- Swell attenuation filter factor

• Perturbed winds vs. Perturbed winds and perturbed physics

• Conclusions

• Future Work

Page 3: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Why the need for a Wave ensemble system?

• Limitations of deterministic models

- Can miss extreme events.

• Obtain reliable probabilities of wave events happening.

- Time window of calm waves important for offshore lifting applications.

• Act as a re-tuning exercise for the

operational model

Page 4: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

NAE

• 24-member ensemble designed for short-range forecasting• Regional ensemble over N. Atlantic

and Europe (NAE) (24km resolution) to T+54

• Global ensemble (~90km resolution) to T+60

• ETKF for initial condition pert

• Stochastic physics

• Global run at 0Z and 12Z. Regional run at 6Z & 18Z

• Global creates boundary conditions for the NAE

Wave ensemble system

Page 5: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

WAVEWATCH3 – The Met Office wave model

• Spectral model (24 directions X 25 frequencies resolution)

• Tolman-Chalikov source terms

• Restart each ensemble member from the previous cycles’ forecast (T+12) => Maintain spread at low lead times. esp. swell waves

Page 6: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Introductionperturbed physics scheme

Page 7: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Uncertainties• Uncertainties in the model:

• Initial conditions (Wind forcing, waves)

• Model structure – grid size

• Parameterization of physics

• Previous work investigating the initial conditions have been able to account for 50% of the uncertainty in the wave ensemble system. (Jan 09-Apr 09)

• Adding a perturbed physics scheme aims to increase this and capture more uncertainty.

• Assess over a 3 month period. (Sep 09-Nov 09)

50%

Page 8: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

What is a perturbed physics scheme?• Physics stay the same but the parameter value

varies.

• Fine tuning of the processes in WW3 occur through ‘tuning knobs’ to avoid unbalancing input and dissipation terms.

• Processes occur on scales to small to be resolved e.g dissipation => the need to parameterize

• Leads to uncertainty associated with empirical parameters.

Page 9: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Perturbed physics scheme

• Indentify the key parameters in WW3 likely to cause uncertainty – Lit review (Tolman, 2002) and ‘expert’ discussions,

- Effective wind stress, Ue – (STABSH) - Swell attenuation (SWELLF) - Non-linear interaction terms? - Dissipation?

• Treat a selected group of parameters as stochastic variables (Random sampling from uniform distribution)

• Chosen from appropriate limits

Page 10: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Sensitivity studyEffective wind stress

Page 11: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Effective wind stress (Ue)• Ue governs a more accurate representation of

how wind energy is transferred to waves in the presence of different atmospheric conditions (temperature and moisture stratification)

• c0 (STABSH) assigned values of 1.3, 1.35, 1.4, 1.45 and 1.5 (default is 1.38). Tolman (2002) originally investigated 1.35, 1.38, 1.4 and 1.42

• Ran for a month with fixed winds.

STABSH

Page 12: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

NAE buoy locations (ndbc)

Page 13: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

North west of Scotland 7/09/09

STABSH = 1.5

STABSH = 1.3

.

.

Observations

Page 14: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

Page 15: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

13/09/09 Track 1203

Hs spread has Non-linear relation with wind speed, possibly related to wind direction as well

Page 16: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

Page 17: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

Page 18: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

Page 19: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

Page 20: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

Page 21: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

STABSH values

RMSE (m) Bias (m) Corr (r)

1.3 0.495 0.040 0.911

1.35 0.515 0.109 0.913

1.4 0.549 0.175 0.912

1.45 0.589 0.244 0.913

1.5 0.637 0.312 0.913

Default is 1.38. A lower value (1.3) gives better statistics against observations for this time period

Page 22: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

STABSH

Sep 09 – NAE domain

15%

Page 23: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Random STABSHPerturb every 12 hours in the forecast length

winds still fixed

7.5%

Page 24: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Sensitivity studySwell attenuation

Page 25: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Swell attenuation filter factor (Xs)

• Swell dissipation due to opposing or weak winds is improved using a filtered input source term:

• Xs was assigned values of 0.1, 0.105 and 0.11 (default is 0.1). Tolman (2002) originally investigated 0.087, 0.1, 0.11 and 0.125

• Ran for a month with fixed winds.

SWELLF

Page 26: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

SWELLF

Page 27: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

SWELLF – 4bin output

• Small effect on Hs, does it effect wave period?

• 4bin gives the energy present in different wave systems and partitions waves into waves periods of 0-5s, 5-10s, 10-15s and 15-20s (level 1). Information can be lost when investigating Hs

• Very Localised change in time and space.

Page 28: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

4bin difference

19/09/09

Lv. 2. (10-15s)

NAE domain

Very localised change in time and space.

Influence on this part of the wave field are still small

Wave height

SWELLF = 1.1

Wave height

SWELLF = 1.11

Middle – top plot

-0.1m0.1m

Page 29: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Perturbed winds vs. Perturbed winds and perturbed physics

Page 30: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

RMSE plots

Perturbed winds Perturbed winds and physics

1/09/09 – 27/09/09 Increase spread of ~5%

Page 31: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Talagrand diagrams (Hs)

Perturbed winds Perturbed winds and physics

Underspread – No clear improvement with perturbed physics scheme

Only one month worth of data

Page 32: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Talagrand diagram (Wind speed)

Perturbed winds run (same for perturbed winds and physics)

still underspread in the system

Page 33: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Spatial Hs difference 19/09

Perturbed winds Perturbed winds and physics

Mean

Spread

Page 34: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Spread-skill diagrams

Perturbed winds Perturbed winds and physics

Only valid up to a spread of ~0.5m.

Little difference between the two runs

Page 35: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Conclusions• Increase in spread with the perturbed physics added is

not as large as expected.

• Randomly changing a physical parameter onto the wind members has a tendency to lose the increase in spread caused by the wind speed.

STABSH and wind intensity are not correlated.

• Interaction between parameters values – Trade of between improving negative biases in the tropics and positive biases in the storm tracks (changes in STABSH and SWELLF are interlinked) (Tolman , 2002)

• A slightly lower STABSH value provided better statistics for Hs in the NAE domain. However, can’t be directly compared to operational model due to resolution differences

• Strong test bed for further studies

Page 36: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Future work• Gain a better understanding of STABSH and SWELLF –

how changes affect the wave field in space and time.

• 3 month validation. • Use satellite data in the validation process (RMSE)

• Possibility of Perturbing more physical parameters.

• Compare to a ‘dressed ensemble’ – bias and std of model given from comparisons to observations at site specific locations. Pros and cons of each approach.

Page 37: Investigating a perturbed physics scheme in a wave ensemble system

© Crown copyright Met Office

Any Questions ???

References

Bocquet, Francois-Xavier, Sauler, A. and Bunney, C. (2009) Wave ensemble scoping study.

Bowler, N.E., Arribas, A., Mylne, K.R., Robertson, B.K., and Beare, S.E. (2008) The MOGREPS short-range ensemble prediction system. Q. J. Meteorol. Soc. 134, 703-722. pp. 703-722.

Tolman, H.L. (2002) Testing of WAVEWATCH III version 2.22 in NCEP’s NWW3 ocean wave model suite. Technical note. July 2002.

Tolman, H.L. and Chalikov, D. (1996) Source terms in a third-generation wind wave model. J. of Phys. Oc. 26, pp. 2497-2518


Recommended