Lead Session Chair:
Stephan Barth, Managing Director, ForWind - Center for Wind Energy Research, Germany
Susumu Shimada (1) F P Teruo Ohsawa (2) Yuko Takeyama (1) Tetsuya Kogaki (1) Gerald Steinfeld (3) Detlev Heinemann (3)
(1) AIST, Koriyama, Japan (2) Kobe University, Kobe, Japan (3) Forwind, Oldenburg, Germany
Printer friendly version: Print
Download poster(1.96 MB)
Presenter's biographyBiographies are supplied directly by presenters at OFFSHORE 2015 and are published here unedited
Dr. Shimada has been working in the renewable energy research field for almost 10 years. He is a currently a research scientist at AIST, which is a national research institute in Japan. He studied at Kobe University as a Ph.D. student in Japan. After he studied he spent two years at Gifu University for a national project of PV forecasting. His research is now focused on offshore wind resource assessments using numerical models.
Comparison of wind speeds simulated with WRF using seven planetary boundary layer schemes at two offshore met masts in the North Sea
The authors conducted offshore wind simulations using a mesoscale model, Weather Research and Forecasting (WRF), developed by National Center for Atmospheric Research (NCAR) and National Centers for Environmental Prediction (NCEP). For finding the optimum configuration for offshore wind resource assessments with WRF, simulations using seven different Planetary Boundary Layer (PBL) schemes are performed. The accuracies are examined by comparing the simulated winds with in situ observations at the two offshore met masts of FINO1 and the Offshore Wind farm Egmond aan Zee (OWEZ), located at 45 and 18 km away from their nearest coastlines in the North Sea.
Accuracy verifications of the WRF simulations at two different offshore met masts are conducted, since it might be a difficult to draw a general conclusion from the results at only one observational station. We employ WRF version 3.5 using three domains, which have horizontal resolutions of 18, 6 and 2 km, respectively. Vertically, 40 layers are configured from the surface to the 50 hPa pressure level. The WRF hindcast for July 2006 are run using first order turbulent closure schemes (ACM2, MRF and YSU) as well as Turbulent Kinetic Energy (TKE) based schemes (MYJ, MYNN2, MYNN3 and QNSE).
Main body of abstract
Since the wind speeds within the PBL in the WRF model are impacted by the choice of the PBL scheme and its associated surface layer scheme, the significance of the PBL scheme choice has been well understood. However, it is still obscure which PBL scheme is the best for offshore wind simulations using WRF. For instance, Muñoz-Esparza et al. (2012) compared the performance of five PBL schemes at FINO1 and they indicated that the MYNN2 scheme was the most suitable scheme. Also, Draxl et al. (2014) evaluated the winds and vertical wind shear from WRF simulations with seven PBL schemes at an onshore met mast in Danish coastal water and they mentioned that the MYJ PBL scheme case showed the best performance.
The bias, Root-Mean-Square Error (RMSE) and correlation coefficient in the WRF wind speeds are analyzed at FINO1 and OWEZ. The accuracy verifications are conducted at several heights for different stability classes. The run with the MYJ scheme has smaller biases and RMSEs, and larger correlation coefficients than the runs with other PBL schemes at FINO1 for almost all heights and stability classes. Similarly, the MYJ scheme case showed better scores in bias for the most stability cases, although the other PBL schemes have better scores in RMSE and correlation coefficient in some cases at OWEZ. Consequently, the MYJ scheme appears to be the most promising scheme and the results are more in line with the conclusions in Draxl et al. (2014) than those in Muñoz-Esparza et al. (2012).
All in all, the WRF model showed a good performance for offshore wind simulations at the two offshore sites. However, some issues that need to be enhanced still remains. For instance, comparisons of the accuracies between the sea and land fetches at OWEZ showed that the accuracy of WRF obviously decreases when the wind came from land to sea. Thus, we think that some improvement is still needed for accurately assessing wind resources near the coast by the WRF model, although the simulated results without unreasonable biases can be expected for sea fetches by the WRF offshore wind simulations.
The WRF model is the most widely used meteorological model for regional scale site assessments. Thus, the results in this study could be of interest to people who are working on wind resource assessments using numerical models. Moreover, since the strong and weak points of the numerical model simulation for offshore wind resource assessments will be discussed, the presentation might be useful not only for the WRF users but also offshore wind power plant developers.