1. 1 An operational wavemodel for the Faroe Shelf Thesis defence 2/4-2007 Bárður A. Niclasen Bárður A. Niclasen Statoil, Phillips, Enterprise og Veba Náttúruvísindadeildin

2. 2 Tidal currents and their influence on sailing conditiones

3. 3 Effect of stationary currents on waves (size<20 km) (quasi-stationar approximation) Wave height and steepness increase while absolute period remains constant Wave Current

4. 4 Large breaking wave hit Sjóberin at ~18:30 on the 8/3-2007 Reason : current treveled in opposite direction of the wind

5. 5 Buoy Time H m0 Dir Buoy Time H m0 Dir North WV-312:04 9.7 338 West WV-2 12:22 5.4 359 East WV-111:51 5.6 17 South WVD-412:39 7.4 342 Existing operational wave forecast http://vs.en.sigling.is – based on ECMWF forecast http://ocean.dmi.dk

6. 6 Motivation for this project The impact of tidal currents on the sea state in the coastal sone of Faroe Islands are very important...but these effests are not included in present operational wave models for the area Need local wave model with increased resolution Need local wave model with increased resolution that includes the effect from the tidal currents on the waves. that includes the effect from the tidal currents on the waves. As this can improve sea safety in the long run As this can improve sea safety in the long run

7. Outline General information General information Validation of wave forecasts from ECMWF Validation of wave forecasts from ECMWF Optimal settings of a local wave model (SWAN) Optimal settings of a local wave model (SWAN) SWAN runs including the effect from tidal currents on the sea state. SWAN runs including the effect from tidal currents on the sea state. Conclusions/Outlook Conclusions/Outlook

8. 8 Waves according to linear theory Steepness of the wave is H/λ

9. 9 Wave parameters: H m0 = 2.0m wave height T p = 6.3s peak-period T m02 = 4.9s average-period... Wave measurement example T -1 H m0 =2.0 m T p =6.3 s

10. 10 How linear wave models represent real sea states

11. 11 Governing equations in wave model Propagation: where, Dynamics:

12. 12 ECMWF validation 4 hindcast periods spanning one month 4 hindcast periods spanning one month All include one major storm (the 4 largest in 1999-2004 at WVD-4) All include one major storm (the 4 largest in 1999-2004 at WVD-4) 2 summer events and 2 winter events 2 summer events and 2 winter events EventPeriod 1 20/8 - 20/9 2000 2 15/1 - 15/2 2002 3 1/6 - 1/7 2002 1/6 - 1/7 2002 4 1/1 - 1/2 2003 1/1 - 1/2 2003 WV-1, WV-2, WV-3 & F-8: Landsverk. WVD-4: DataQuality. K7: MetOffice.

13. 13 Time series from WVD-4 vs. ECMWF in Event 1

14. 14 Wave spectra at WVD-4. in Event 1 Model Measured

15. 15 Statistical average from all events SiteParameterBiassummer/winterSc.Index WVD-4 H m0 0.0/-0.70.18 - T m02 0.50.13 - TpTpTpTp0.10.16 WV-2 H m0 0.0/-0.30.20 - T m02 0.60.19 K7 H m0 -0.1/-0.70.18 - T m02 -0.20.09

16. 16 Effect of high ferquency cut off Cut-off frequency f high [Hz] 0.580.550.500.450.400.350.300.250.200.150.10 H m0 Bi Sc 0.00 -0.01 0.00 -0.01 0.00 -0.01 0.01 -0.02 0.01 -0.03 0.01 -0.05 0.02 -0.09 0.03 -0.17 0.07 -0.39 0.15 -1.16 0.44 T m0-1 Bi Sc 0.03 0.00 0.04 0.01 0.06 0.01 0.08 0.01 0.12 0.02 0.17 0.03 0.26 0.04 0.42 0.06 0.72 0.10 1.39 0.18 3.49 0.44 T m01 Bi Sc 0.13 0.02 0.15 0.02 0.20 0.03 0.26 0.04 0.35 0.05 0.47 0.07 0.64 0.10 0.92 0.14 1.37 0.20 2.24 0.32 4.50 0.64 T m02 Bi Sc 0.30 0.05 0.34 0.06 0.43 0.07 0.53 0.09 0.66 0.11 0.84 0.14 1.08 0.18 1.43 0.24 1.98 0.32 2.96 0.48 5.31 0.85 WVD-4 f high = 0.58 Hz, WV-2 f high = 0.50 Hz, K7 f high = 0.25 Hz padded f -5 tail measurements T m02 bias WVD-4: 0.5 s WV-2: 0.6s K7: -0.2s After correcting for different measurement range (padding f -5 tail) WVD-4: 0.8 s WV-2: 1.0s K7: 1.2s

17. 17 Conclusions – on ECMWF validation Overall it is found that ECMWF analysis wind and wave data are suited to force local wave model Overall it is found that ECMWF analysis wind and wave data are suited to force local wave model WV-1 and WV-3 not suited for validation WV-1 and WV-3 not suited for validation Seasonal variation with less negative H m0 bias in summer periods compared to winter periods Seasonal variation with less negative H m0 bias in summer periods compared to winter periods Smooth predicted spectra with some missing swell events Smooth predicted spectra with some missing swell events Slightly less energy in spectral tail compared to measurements (positive bias in T m02 ) Slightly less energy in spectral tail compared to measurements (positive bias in T m02 )

18. 18 Choosing a local model ModelNumericsCurrentsTuning WAM4expl.stat.none WW3expl.stat./unstatrequiered SWANimpl.stat./unstatsome SWAN was choosen because: SWAN was choosen because: High resolution runs favour implicit propagation High resolution runs favour implicit propagation Want to include unstationary currents Want to include unstationary currents Less site specific tuning Less site specific tuning

19. 19 Nested SWAN model domains Nesting 3 1/60 ° x 1/120 ° 72 directions dt=15 min Nesting 1 1/8 ° x 1/16 ° 24 directions dt=30 min Nesting 2 1/60 ° x 1/120 ° 24 directions dt=30/15 min

20. 20 Different model physics/numerics Dynamics Dynamics Run type Comments1G 1. generation physics 2G 2. generation physics Komen 3G WAM3 physics Komen n=... 3G WAM3 physics with changed n in S ds Janssen 3G WAM4 physics Janssen C ds =... 3G WAM4 physics with changed C ds in Sds Janssen-Roop 3G WAM4 physics with WAM4.5 lmiter Westhuysen 3G New

21. 21 Different SWAN runs, Event 1, Nesting 1

22. 22 Intermediate results from Nesting 1 SWAN-Komen: H m0 and T p OK, but serious negative bias in T m02 SWAN-Komen: H m0 and T p OK, but serious negative bias in T m02 SWAN-Janssen: T p and T m02 OK, but negative bias in H m0 SWAN-Janssen: T p and T m02 OK, but negative bias in H m0 SWAN-1G and SWAN-2G: OK, but too much scatter in results SWAN-1G and SWAN-2G: OK, but too much scatter in results A need to retune source terms in SWAN A need to retune source terms in SWAN

23. 23 Whitecapping dissipation in SWAN SWAN-Komen (WAM3) : SWAN-Janssen (WAM4):

24. 24 S ds,w n-tuning no f -5 tail with f -5 tail Least error if n=2.0-2.1 Least error if n=1.9

25. 25 Time averaged spectra from modelled spectra vs. data from WVD-4 in Event 1 ECMWF-averaged spectra close to the measured, but slight undershoot in the high frequencies Default SWAN poor fit to the measured spectra, but good fit with new n Default SWAN- Janssen poor fit to the measured spectra, but good fit with new limiter

26. 26 Compatability at the boundary Waves Do the nested models accept the incoming sea state? If the nested models are compatible with the forcing they should have H m0 and T m02 close to the forcing model (black) Default SWAN is not compatible with the forcing model, but retuned SWAN- n=2 and SWAN- Janssen with new limiter are compatible with the ECMWF-WAM4 forcing !!

27. 27 Conclusions from 1.-nestings Optimal nested-model performance with SWAN- Komen with n =1.9 (2.0-2.1) Optimal nested-model performance with SWAN- Komen with n =1.9 (2.0-2.1) SWAN-Komen with retuned n much better compatibility with ECMWF-WAM4 than default setup SWAN-Komen with retuned n much better compatibility with ECMWF-WAM4 than default setup Hersbach-Janssen limiter greatly improves the SWAN- Janssen runs, but runs are unstable in high resolution Hersbach-Janssen limiter greatly improves the SWAN- Janssen runs, but runs are unstable in high resolution

28. 28 2. and 3.-nestings with tidal currents Direct validation of the numerical tidal model not possible, but it can be compared against independent tidal predictions based on measurements One such example from a location close to the south buoy... U-component (East) of the tidal current ellipse in [m/s] for Event 1

29. 29 Influence from tidal currents on wave height in idealized test case (difference plot) Waves H m0 =2.5m Wind 10 m/s m

30. 30 Influence from tidal currents on wave steepness in idealized test case (difference plot) Offshore wave steepness ~ 0.058 Waves H m0 =2.5m Wind 10 m/s

31. 31 Influence from tidal currents on wave dissepation in idealized test case Waves H m0 =2.5m Wind 10 m/s

32. 32 3.nesting with and without currents example where the tidal effect is clear at WVD-4

33. 33 Conclusions from 2. and 3. nestings SWAN capable of recreating most of the tidally induced variations seen in the measurements SWAN capable of recreating most of the tidally induced variations seen in the measurements SWAN-Komen with n=2 is a better option than default SWAN for deepwater areas with unstationar currents SWAN-Komen with n=2 is a better option than default SWAN for deepwater areas with unstationar currents DIA (S nl ) counteracts variations in spectral shape (T m02 ) forced by unstationar currents DIA (S nl ) counteracts variations in spectral shape (T m02 ) forced by unstationar currents Lag in measured H m0 variations compared to relative current due caused by dynamical effects (up-wind slowing down of wave energy) Lag in measured H m0 variations compared to relative current due caused by dynamical effects (up-wind slowing down of wave energy)

34. 34 Suggested operational setup Reducing computational cost 2.nesting only 2.nesting only 1.-order propagarion 1.-order propagarion Larger time step Larger time step Comp. time v.s realtime Nest 3 setup extended to the entiere nest-2 domain: comp. ratio ~ 1.4 (i.e. 2-day forecast finnished after 2.8 days) Operational setup of nest-2 area: comp. ratio ~ 0.1 i.e. 2-day forecast in 5 hours if serial comp. or approx. 1.0-1.5 hours if paralell comp.)

35. 35 Summary Regional wave/wind model is vaildated for the Faroese area (ECMWF) Regional wave/wind model is vaildated for the Faroese area (ECMWF) Local wave model is implemented, adjusted and verifyed Local wave model is implemented, adjusted and verifyed Operational setup for a local wave model including the effects from tidal currents is suggetsed. Is possible to run on the local linux- closter. Operational setup for a local wave model including the effects from tidal currents is suggetsed. Is possible to run on the local linux- closter.

36. 36 Outlook Operational waveforecasts that include the effect of unstationary tidal currents are possible for the Faroe Shelf... if $$! Operational waveforecasts that include the effect of unstationary tidal currents are possible for the Faroe Shelf... if $$! Implementation of high-resolution in wave- and wind-models, inclusion of wave reflections, dynamic currrents (oceanic, wind, pressure) Implementation of high-resolution in wave- and wind-models, inclusion of wave reflections, dynamic currrents (oceanic, wind, pressure)

37. 37 Thank you

38. 38 Don't be cruel....pity me!!

39. 39 Wave spectra at WVD-4, event 1, 3.-nesting

40. 40 3.nesting with and without currents

41. 41 Intermediate conclusions from 3.nestings Default SWAN-Komen better predictiones without including the effects of the current Default SWAN-Komen better predictiones without including the effects of the current SWAN-Komen with n=2 better predictiones when including the effects from the current SWAN-Komen with n=2 better predictiones when including the effects from the current Variations in wave height and direction recreated to resonable extent when including the influence from the current Variations in wave height and direction recreated to resonable extent when including the influence from the current Modelled variations in T m02 are missing... ? Modelled variations in T m02 are missing... ? Modelled variations in H m0 are lagging... ? Modelled variations in H m0 are lagging... ?

42. 42 3.nestings with different source terms turned off

43. 43 Numerical tidal model used in the 2. and 3.-nestings model vs. tidal ( prediction from measurements at WV-4) Estimated maxmum current strength

44. 44 Effect of stationary currents on waves (size<20 km) (quasi-stationar approximation) Wave height and steepness change while absolute period remains constant