PhD-stud. Olav Krogsæter olav.krogsaeter@stormgeo.com

Dr. Knut Lisæterknut.lisaeter@stormgeo.com

New two-way coupled atmosphere-wave model system for improved wind

speed and wave height forecasts for offshore wind energy applications

Outline

● WRF, SWAN, and the coupled system

● Results● Three cases:

– Stormy weather.

– Cold air outbreak.

– Inversion.

● Yearly statistics.

● Summary

WRF Model

Non-hydrostatic mesoscale weather prediction system

Weather Research and Forecasting model

Large and growing set of parameterization options

Surface layer schemes

Boundary layer schemes

Microphysics schemes

Cumulus schemes

Radiation schemes

Nesting capability

Nudging capability

Assimilation capability

Open Source project

The SWAN wave model • Simulating WAves Nearshore • Simulates the wave spectrum • Includes effects such as

• Shoaling • Refraction • Whitecapping • Bottom friction

• Has been modified at StormGeo to read 2D-spectra from Grib files

• Run operationally for N. Europe

Coupled model

● The most difficult part was how the SWAN model should influence the WRF model.

● Parameterizing the effect that the ocean surface has on the atmosphere is still an active field of research.

● The key parameter the SWAN model modifies is the roughness length, z0, seen by the WRF model. This is communicated through the Charnock parameter, zch:

z0 = zch (u*)2/g

where u* is the friction velocity and g the gravitational constant.

Coupled model

Stand alone WRF: Charnock parameter is a constant.

Coupled WRF-SWAN model:

i) HEXOS parameterization: The Charnock parameter depends on wave age.

* Developing waves: Increasing roughness with wave age.* Swell: Decreasing roughness with wave age.

---> Charnock parameter becomes a variable

ii) Janson parameterization: The Charnock parameter is a function of wave growth.

Coupled model

• Technical work is done – WRF and SWAN are set up to run within Earth System Modelling Framework, ESMF

• Information exchanged every hour – SWAN receives 10 m winds from WRF – WRF receives a new roughness parameter, (z0), from SWAN

• One year run with both the MYNN2 and MYJ PlanetaryBoundary Layer (PBL) scheme in WRF, coupled with SWAN and the HEXOS parameterization is finished.

–

Coupled model

• Technical work is done – WRF and SWAN are set up to run within Earth System Modelling Framework, ESMF

• Information exchanged every hour – SWAN receives 10 m winds from WRF – WRF receives a new roughness parameter, (z0), from SWAN

• One year run with both the MYNN2 and MYJ PlanetaryBoundary Layer (PBL) scheme in WRF, coupled with SWAN and the HEXOS parameterization is finished.

–

WRF and SWAN: coupled run

+1 hour

WRF

SWAN

WRF and SWAN: coupled run

+1 hour

WRF

SWAN

ESMF

wind

z0

WRF and SWAN: coupled run

+1 hour

WRF

SWAN

ESMF

wind

z0

WRF and SWAN: coupled run

+1 hour +2 hours

WRF WRF

SWAN SWAN

WRF and SWAN: coupled run

+1 hour +2 hours

WRF WRF

SWAN SWAN

ESMF

wind

z0

WRF and SWAN: coupled run

+1 hour +2 hours +3 hours +4 hours

WRF WRF WRF WRF

SWAN SWAN SWAN SWAN

Result: Stormy case in November 2010

Wind Speed uncoupled, 40m Difference wind speed, 40m, uncoupled-coupled

Result: Stormy case in November 2010

Wind Speed uncoupled, 40m Friction velocity, u*

Result: Stormy case in November 2010

Friction velocity, u*, uncoupled Difference, u*, coupled-uncoupled

Result: Stormy case in November 2010

Friction velocity, u* Wind speed, 40 m.

Result: Stormy case in November 2010

Vertical profiles Vertical profiles

Result: Cold air outbreak, March 2010

Wind Speed uncoupled, 40m Difference wind speed 40m, uncoupled-coupled

Result: Cold air outbreak, March 2010

Difference, u*, coupled-uncoupled, MYNN2 Difference, u*, coupled-uncoupled, MYJ

Result: Inversion, SST=18C, Ta=25C, July 2010

Wind Speed uncoupled, 40m Difference wind speed 40m, uncoupled-coupled

Result: Inversion, SST=18C, Ta=25C, July 2010

Friction velocity, u* Wind speed, 40 m.

Result: Inversion, SST=18C, Ta=25C, July 2010

Vertical profiles Vertical profiles

Result: qq-plot, wind speed, summary 2010

Result: qq-plot, u* , summary 2010

Result: 40 m wind speed, summary 2010

2010 Min. 1st Qu Median Mean 3rd Qu Max Mean error

Mean absolute error

Standard deviation

Fino1 0,18 5,60 8,01 8,30 10,58 25,33 3,85

MYNN 0,30 5,79 8,15 8,48 10,75 22,88 0,18 1,44 3,79

MYNN-SWAN

0,34 5,78 8,04 8,32 10,56 21,51 0,02 1,41 3,62

MYJ 0,35 5,90 8,11 8,42 10,53 24,01 0,12 1,46 3,69

MYJ-SWAN

0,33 5,87 8,06 8,28 10,32 23,08 -0,02 1,48 3,53

Summary

●A new two-way coupled atmosphere-wave research and forecasting system is implemented: WRF-SWAN.●Two different PBL-schemes: MYJ and MYNN2.●HEXOS parameterization for computing the new roughness parameter from SWAN that goes into WRF. Function of wave age.●Janson parameterization – ongoing work. Function of wave growth.*****************************************************************************

●Reduces the well-known positive bias in WRF with both PBL-schemes.●Reduces the MAE in the MYNN2-SWAN setup.●Increases slightly the MAE in the MYJ-SWAN setup. ●Strong winds greater than 15 m/s are reduced too much in the coupled runs.****************************************************************************

●From previous research on many different PBL-schemes by e.g. O. Krogsæter (2013) and A. Hahmann (2012):

* MYJ scheme perform best in offshore conditions with WRF stand alone.

* MYNN2 scheme perform slightly better in this new coupled system.

Thank you!