21 April 2010 EWEC 2010 Warsaw Jan Coelingh & Anthony Crockford

1. 21 April 2010 EWEC 2010 Warsaw Jan Coelingh & Anthony Crockford Best Practice for use of mesoscale models in wind resource assessments

2. Outline • Background • Wind atlas products • Methods • Validation in North Sea • Conclusions

3. Background • What is mesoscale modeling? • Mathematical models of the atmosphere that predict the weather, using observed data from a variety of sources, such as weather satellites and surface weather observations • Horizontal grid spacing in the model is typically between about 1 km and 20 km • Many processes are approximated through parametrisation

4. Mesocale-based wind atlas (Anemos)

5. How useful are mesoscale model results for wind resource assessments at sites without measurements? • Can a mesoscale wind atlas be used as a stand-alone product? • How should a mesoscale wind atlas be used in conjunction with measured wind data? • Validation exercise using two methods on two mesoscale model products

6. Method A – Direct application • Take a time series from mesoscale wind atlas at nearest grid point to the site • Describe terrain around grid point using the same basis as in the mesoscale modelling • Calculate generalised local wind climate, using the WAsP wind atlas methodology The result of this post-processing is a regional wind statistic containing wind speed parameters for standard heights and roughness classes

7. Method B – Correction factor applied to distant measurements • Calculate generalised local wind climate, based on measurements • Take a time series from mesoscale wind atlas at nearest grid point to the site, and nearest grid point to measurement site • Calculate correction factor (mean wind speed near site / mean wind speed near measurement site) • Apply correction factor to wind climate based on measurements

8. Validation exercise in North Sea

9. Validation exercise in North Sea • Five data sets from KNMI points in The Netherlands • Europlatform 29 m 45 km • K13 74 m 105 km • Meetpost Noordwijk 28 m 10 km • L.E. Goeree 38 m 17 km • Vlakte van de Raan 17 m 15 km • One data set from MVB point in Belgium • Westhinder 25 m 30 km • Dedicated met mast in The Netherlands • OWEZ 70 m 15 km • Dedicated met platform in Germany • FINO 100 m 50 km • Dedicated met mast in southern North Sea • CONFIDENTIAL 40 m 6 km

10. Validation exercise in North Sea

11. Validation exercise: Method A

12. Validation exercise: Method A

13. Validation exercise: Method B

14. Validation exercise: Method B

15. Further validation exercises • We have also completed a number of validation exercises for onshore sites in: • France • Poland • UK • With careful use of ‘Method A,’ wind farm energy yield is predicted to within ± 3-12%

16. Conclusions (1) • Be aware of mesoscale model conditions: source data, boundaries, resolution, time scale • Select site-appropriate mesoscale models • e.g. higher resolution for coastal or complex sites • Validate mesoscale model results at several nearby sites

17. Conclusions (2) • With proper post-processing, mesoscale wind atlases are suitable for preliminary WRAs • Combination of mesoscale wind atlas and micro-scale modeling allows for effective site assessments prior to measurements • Mesoscale model results can complement analysis based on distant measurements • The added-value of wind-speed gradients is especially evident offshore • Recommendation: make estimate of uncertainty and clearly state it

18. Acknowledgements • Data Europlatform, Meetpost Noordwijk, L.E. Goeree, Vlakte van de Raan courtesy of KNMI • Data OWEZ courtesy of Noordzeewind • Data FINO 1 courtesy of German government • Mesoscale map courtesy of Anemos