The selective dynamical downscaling method for extreme wind atlases

1. The selectivedynamicaldownscalingmethodfor extremewindatlases Xiaoli Guo Larsén Jake Badger Andrea N. Hahmann Søren Ott EWEC 2011

2. Why is such a method needed? • Lack of long term measurements • Global model data: not sufficient in resolution • Regional climatemodeled data: not optimal in storm events EWEC 2011

3. The method • 3 steps of the selectivedynamicaldownscalingmethod : • Identification of storm events for a selectedarea • Mesoscalemodelingusing WRF • Post-processing EWEC 2011

4. The method Step 1: Identify storm events for a selectedarea basedon -- annualwindmaximummethod for U50 -- annual max G and u10 EWEC 2011

5. The method Step 2: Mesoscalemodelingusing WRF (Denmark case) • Run WRF for the 58 cases • WRF setup: • WRF V3.1 • FNL 1 data, 6 hrly • SST 0.5 • 45 – 15 – 5 km • 37 verticallayers • YSU PBL scheme • Liu et al microphysics • Time step: 4 min (D1) • Run time <=48 hrs, nonudging • 10 min output • The 50-year windusing the AnnualMaximaMethod. EWEC 2011

6. The method • Step 3: Post-processing • To generalize the winds to standard conditions(at a certainheight over a homogeneoussurface of certainroughness, here 10 m orhubheight, roughness length=5 cm) • To prepare for data validation • To prepare platform to pass the mesoscalewinds to microscalemodeling EWEC 2011

7. The method • Step 3: Post-processing • For measurements (WAsP cleaning procedure): u0,z : measuredwind at heightz so,sr: sectorwisespeed-upcoefficients for orography and roughnesschange z0= 5 cm z0= 5 cm u*,r ust EWEC 2011

8. The method • Step 3: Post-processing • For WRF output, approach-1: WRF winds Roughness length used in WRF z0= 5 cm z0= 5 cm u*,r ust EWEC 2011

9. The method • Using LINCOM (from Risø) to the orography and roughnessmaps as used in WRF, weget • Coefficients of directionalorographicalchange at differentheights • Coefficients of directionalupstreamroughnesschange at differentheights • Coefficients of directionalupstreamroughness – effectiveroughness • Step 3: Post-processing • For WRF output, approach-2: z0= 5 cm z0= 5 cm u*,r ust Details in Badger et al. 2010: A universal mesoscale to microscalemodeling interface tool. EWEC Warsaw, Poland, 2010) EWEC 2011

10. The method • Step 3: Post-processing • For WRF output, approach-2: e.g. West Sector Upstreamorographychange @50 m Upstreamroughnesschange @ 50 m Effectiveroughnesslength 10 EWEC 2011

11. Results (nopost-processing) The 50-year winds at different model levels, including 10 m, 15 m, 50 m, 105 m U50 @ 10 m 32 m/s Latitude 24 m/s 16 m/s Longitude EWEC 2011

12. Results (nopost-processing) EWEC 2011

13. Results (nopost-processing) * * : due to Gumbel fitting EWEC 2011

14. Results (withpost-processing) The 50-year wind at standard condition withpost-processing approach-1 (PP 1) EWEC 2011

15. Results (withpost-processing) The 50-year wind at standard condition withpost-processing approach-1 (PP 1) 15 EWEC 2011

16. Results (withpost-processing) The 50-year wind at standard condition withpost-processing approach-2 (PP 2) EWEC 2011

17. Results (withpost-processing) The 50-year wind at standard condition withpost-processing approach-2 (PP 2) 17 EWEC 2011

18. Results (withpost-processing) Stations wrf PP1 wrf PP2 OBS ±  The 50-year wind of standard conditions (at 10 m, over z0=0.05 m). Charnock parameter = 0.03. *: values from Larsén and Mann (2009): Extremewinds from NCEP/NCAR reanalysis data, Wind Energy EWEC 2011

19. Conclusions • The selective dynamical downscaling method is robust and efficient • The 50-year winds from this method using WRF simulatiuons without post-processing are reasonable – offshore extreme winds seem underestimated • The 50-year winds corrected to standard conditions are in better agreement with measurements that are also corrected to the same conditions EWEC 2011

20. Conclusions • The selective dynamical downscaling method is robust and efficient • The 50-year winds from this method using WRF simulatiuons without post-processing are reasonable – offshore extreme winds seem underestimated • The 50-year winds corrected to standard conditions are in better agreement with measurements that are also corrected to the same conditions • In places of simple orography and roughness fields, post-processing approaches I and II do not bring significant differences • Otherwise, post-processing approach II takes into account the upstream orographical change, roughness change and effective roughness length, and it gives better spatial distribution than the simpler approach I. • Improvement of WRF simulation is needed over ocean EWEC 2011

21. Conclusions • The selective dynamical downscaling method is robust and efficient • The 50-year winds from this method using WRF simulatiuons without post-processing are reasonable – offshore extreme winds seem underestimated • The 50-year winds corrected to standard conditions are in better agreement with measurements that are also corrected to the same conditions • In places of simple orography and roughness fields, post-processing approaches I and II do not bring significant differences • Otherwise, post-processing approach II takes into account the upstream orographical change, roughness change and effective roughness length, and it gives better spatial distribution than the simpler approach I. • Improvement of WRF simulation is needed over ocean EWEC 2011

22. Thanks for your attention Acknowledgement: Thiswork is supported by Danish PSO grant 2009-1-10240 and EU SafeWindproject (213740) Data from FINO 1 areprovided by DeutschesWindenergieInstitut, German Wind Energy Institute, through EU-Norsewind project. EWEC 2011