Analysis of four decadal simulations of the Skagerrak mesoscale circulation using two ocean models...

Analysis of four decadal simulations of the Skagerrak mesoscale circulation using

two ocean models

Lars Petter Røed1 and Jon Albretsen2

Presented at OPNet, Geilo, Norway, May 27, 20101Also affiliated with Department of Geosciences, University of Oslo, Norway2 Now at Institute of Marine Research, Flødevigen, Arendal, Norway

Norwegian Meteorological Institute met.no

LPR #2 OPNet, Geilo, Norway, May 27, 2010

Outline

• Background and motivation• Models and configuration• Results• Conclusions



The story ....

• Switching ocean forecast model– MIPOM -> ROMS

• Comparison of model results vs. observation

• Heat flux parameterization– Calibrating canonical model version



The research learning loop



Key questions:

• Is the new model better, or at least as good as the old model?

• If it's not, is it easy to fix?• Method:

– Comparison model vs. observations– Requires (long) simulations



Area of interest:

Onsdag 2007-03-28 12UTC

Note mesoscale patterns in SST

Satellite image12 h composite Mar 28, 2007

+48 h forecast valid May 9, 2010 00UTC



Grids employed

1.5 km4 km



Item MIPOM ROMSResolution 4 km 1.5 km 4 km 1.5 km

# of vertical levels 26 21 32 32

Long (internal) time step

150 s 60 s 120 s 90 s

Ratio of internal to external time step

30 40 30 30

Horizontal dissipation Smagorinsky No explicit diffusion

Vertical mixingMellor-Yamada

2.5 levelGLS mixing scheme

Horizontal advection scheme

2nd order centered 3rd order upwind

Surface fluxes MI-IMStandard ROMS bulk

fluxes (analytic)

Model facts



Model facts ...

• Simulation period:– 1981–2007 (+ 1980 spin-up)

• Four simulations– 2 x MIPOM : 4 km and 1.5 km– 2 x ROMS : 4 km and 1.5 km

• Atmospheric forcing: – ERA40 + ECMWF operational analysis

2002-2007

• Tidal forcing:– 8 major tidal components– Forced at open boundaries



Model facts ...

• Other forcing at open boundaries:– 4 km:

• SODA reanalysis + climatology 2005-2007

– 1.5 km:• Nested into 4 km

• Rivers:– Climatology

• Baltic outflow– S=12 psu

• No data-assimilation!



Observations for validation

• IMR, Norway – Current

measurements (one location, 160 day period)

– Monthly data from the Hirtshals – Torungen section (12 stations, all years)

• ICES database– Irregular data in time

and space (approx. 250.000 measurements of S and T in Skagerrak)



Observed surface velocities:

• Courtesy: Rio05, CLS Space Oceanography Division, AVISO

• Speed



Currents off western Norway (50 m)

ROMS 4kmMIPOM 4km

Shetland

1981-2007 average (m/s)

Norway



Comparison with earlier results

LaCasce et al. (2007) concluded:The ROMS velocities are more energetic and compare more favorably with in situ observations, as do the ROMS-derived means

MIPOM ROMS

2004-2006 average speed at 50 m



Surface currents Skagerrak

ROMS 4kmMIPOM 4km

1981-2007 average (m/s)



Surface currents Skagerrak

ROMS 4km

1981-2007 average (m/s)

ROMs 1.5km



Temperature bias (mod – ICES obs)

1981-2007 average; upper 20 m

ROMS 4kmMIPOM 4km




1981-2007 average; 50-600 m

ROMS 4kmMIPOM 4km



Temperature bias: 0-20m

Red shading: Model too warm (ci: 0.5ºC)

ROMS v2 MIPOM ROMS v1



Salinity bias (mod – ICES obs)

1981-2007 average; upper 20 m

ROMS 4kmMIPOM 4km



1981-2007 average; 50-600 m

Salinity bias (mod – ICES obs)

MIPOM 4km ROMS 4km



2005average

1983average

ROMS 4kmMIPOM 4km

T-S diagram

Blue: ICES data from SkagerrakRed: Model

ROMS 4kmMIPOM 4km



Temperature bias (mod – IMR obs)

1981-2007 average

ROMS 4km

0

100

600

MIPOM 4km



Temperature bias (mod – IMR obs)

1981-2007 average

ROMS 1.5km

0

100

600

MIPOM 1.5km



Salinity bias (mod – IMR obs)

1981-2007 average

ROMS 4km

0

100

600

MIPOM 4km



Salinity bias (mod – IMR obs)

1981-2007 average

ROMS 1.5km

0

100

600

MIPOM 1.5km



10,0

11,0

0,07

0,03

2,6

2,0

1,50,3

0,60,1

1,90,2

ROMS 4km

12,7

10,1

0,04

0,03

2,5

2,0

1,40,3

0,50,2

1,40,2

MIPOM 4km

Volume fluxes

1981-2007 average (Sv)

More intense cyclonic circulation in Skagerrak!Net flux through Faroe-Shet is negative!



234

310

1,7

1,4

89

7749

10

217

47

7,1

253

353

3,6

1,4

104

8460

11

266

72

8,8

7,7 9,7

1981-2007 average (TW)

Heat transport

MIPOM 4km ROMS 4km

50% more heat advected in the cyclonic circulation in Skagerrak



458

363

1,3

0,9

90

7351

10

19

6

50 5,5

359

397

2,8

0,9

94

7454

10

215

67 6,4

1981-2007 average (t/s)

Salinity transport

MIPOM 4km ROMS 4km

30% more salt advected in the cyclonic circulation in Skagerrak (re-circulation?)



Scatterdiagram – currents: Any skill?

0 m/s 1

1 m/s13 m

50 m



Validation of currents - speed

4km

1.5km

MIPOM ROMS

Speed PDFs: 160 days (27.10.1992-4.4.1993), 13m depth

58.37N,8.51EMeasured total depth: 120mEquilibrium depth: 4km: 233m 1.5km: 163m



Validation of currents - speed

4km

1.5km

MIPOM ROMS

Speed PDFs: 160 days (27.10.1992-4.4.1993), 75m depth




ConMan: Speed at S1 at 100 m

West Norway

Measurement period 2.5 years



ConMan: Speed at S2 at 100 m

West Norway



Validation of currents - direction

4km

1.5km

MIPOM ROMS

Direction PDFs: 160 days (27.10.1992-4.4.1993), 13m depth


Currents from the NE parallel to the:- coast: 238 deg- local isobaths: 225 deg



Validation of currents - direction

4km

1.5km

MIPOM ROMS

Direction PDFs: 160 days (27.10.1992-4.4.1993), 75m depth


Currents from the NE parallel to the:- coast: 238 deg- local isobaths: 225 deg



Validation of currents

160 days (27.10.1992-4.4.1993)

58.37N,8.51EMeasured total depth: 120mEquilibrium depth: 4km: 233m

1.5km: 163m

Mean Standard deviation



ROMS 4kmMIPOM 4km

above 100 m

below 100 m

Obs

Model

0,12

0,35

0

0

Temperature probability distribution

Fre

qu

en

cy

1981-2007 average

MIPOM 4km ROMS 4km

-5 20



ROMS 1.5kmMIPOM 1.5km

above 100 m

below 100 m

Obs

Model

0,12

0,35

0

0

Temperature probability distribution

Fre

qu

en

cy

1981-2007 average

MIPOM 1.5km ROMS 1.5km

-5 20



ROMS 4kmMIPOM 4km

above 100 m

below 100 m

Obs

Model

0,2

0,7

0

0

Salinity probability distribution

Fre

qu

en

cy

1981-2007 average

MIPOM 4km ROMS 4km

24 36



ROMS 1.5kmMIPOM 1.5km

above 100 m

below 100 m

Obs

Model

0,2

0,7

0

0

Salinity probability distribution

Fre

qu

en

cy

1981-2007 average


24 36



20,00

22,00

24,00

26,00

28,00

30,00

32,00

34,00

19

80

19

85

19

90

19

95

20

00

20

05

Sal

init

y

Observed MIPOM ROMS

SSS off southern Norway



0,00

5,00

10,00

15,00

20,00

19

80

19

85

19

90

19

95

20

00

20

05

Te

mp

era

ture

(C

)

Observed MIPOM ROMS

SST off southern Norway



Mean kinetic energy

1981-2007 average (J/m2)

MIPOM 4km ROMS 4km8220

7062

13008 13360

7173

10655

20055 17547



Mean kinetic energy

1981-2007 average (J/m2)


12012 1539818558



Eddy kinetic energy

1981-2007 average (J/m2)

MIPOM 4km ROMS 4km



Eddy kinetic energy

1981-2007 average (J/m2)




Horizontal shear conversion

1981-2007 average (W/m2)

MIPOM 4km ROMS 4km

Regions with negative (red colours) convert energy into eddy kinetic energy



Horizontal shear conversion

1981-2007 average (W/m2)





Vertical shear conversion

1981-2007 average (W/m2)

MIPOM 4km ROMS 4km




Vertical shear conversion

1981-2007 average (W/m2)





Baroclinic conversion (sum of all sub-terms)

MIPOM 4km ROMS 4km

Regions with negative (blue colours) convert energy into eddy kinetic energy

Dominant terms



Baroclinic conversion (sum of all sub-terms)



Dominant terms



Baroclinic conversion (sub-term no. 1)

MIPOM 4km ROMS 4km





MIPOM 4km ROMS 4km




Conclusions

• Overall new model (ROMS) performs better than the old model (MIPOM) – Better current frequency distributions– Slightly better skill

• Surface flux parameterization important – Large impact on the heat content– Modified ROMS version superior in both

temperature and salinity distributions

LaCasce et al. (2007) concluded:The ROMS velocities are more energetic and compare more favorably with in situ observations, as do the ROMS-derived means



Final Remarks

• Increased resolution (4km -> 1.5km):• Better score on validation of currents due

to better resolved topography• Validation score in S and T slightly

worsened

• MKE patterns• Comparable, but there are differences

• EKE patterns• Not comparable

• Baroclinic energy conversion:• Due to the pulsating salinity front• Very clearly seen along the Norwegian

Coastal Current



J. Albretsen and

L. P. Røed

JonsMod 2008




1981-2007 average; 50-600 m

ROMS 4kmMIPOM 4km

Analysis of four decadal simulations of the Skagerrak mesoscale circulation using two ocean models...

Documents

Transcript of Analysis of four decadal simulations of the Skagerrak mesoscale circulation using two ocean models...