5:15 PM - 7:15 PM
[MAG32-P05] Recent advances in satellite SAR for wind energy applications
Keywords:Synthetic aperture radar, SAR, Ocean wind speed, Wind energy
Offshore wind farm planning based on satellite Synthetic Aperture Radar (SAR) has been developed and improved for around two decades. The monitoring SAR satellites, such as the European Copernicus Sentinel-1A/B/C series, have recently been supplemented with small SAR satellites. Small SAR satellite constellations can frequently observe an area of interest and provide high spatial resolution. Sentinel-1 carries C-band SAR, while the new small satellites carry X-band SAR.
An investigation on the accuracy of SAR wind speeds derived from the Japanese StriX showed comparable results to Sentinel-1 using meteorological observations from a mast in the North Sea. Wind speed from StriX vs. mast data showed RMSE 1.38 m/s and bias -0.42 m/s for 41 collocated samples, while wind speed from Sentinel-1 vs. mast data showed RMSE 1.83 m/s and bias -1.02 m/s for 838 collocated samples (Badger et al., 2023). The geophysical model function (GMF) applied was X-MOD2, developed for the German TerraSAR-X. A new GMF may improve wind speed retrieval.
StriX observed winds over an offshore wind farm in the North Sea, and wake conditions from StriX and Sentinel-1 showed a speed-up effect downwind in 48% and 34% of cases, respectively (Hasager et al., 2024). The result is remarkable, as a common understanding is that the wind speeds are lower downstream of a wind farm due to the wake. At hub-height, wind speeds are lower, but at the 10 m height of SAR wind speed, the mesoscale model applied in the study supported the finding of speed-up (Hasager et al., 2024).
Another small Japanese satellite, ASNARO-2, carries an X-band SAR. It was found that neither X-MOD2 nor an improved version of X-MOD2 suitable for the Italian COSMO-SkyMed was suitable for ASNARO-2. A new GMF for ASNARO-2 was developed using numerical weather model wind speed. The ASNARO-2 wind speed results showed RMSE 1.37 m/s and bias -0.10 m/s for 8128 collocated samples from the weather model, and comparison results RMSE 1.68 m/s and bias 0.03 m/s for 22 collocated samples from ocean buoy data (Takeyama & Kurokawa, 2024).
A major advantage of combining wind speeds from several SAR satellites for wind resource assessment is that the collection will be more frequent and at more times during the day. The latter is particularly relevant for offshore areas with pronounced diurnal wind speed variations, such as land-sea breezes (Badger et al., 2023).
Efforts to improve wind speed retrieval from Sentinel-1 C-band SAR were undertaken using three different established numerical models, namely the Global Forecasting System (GFS), re-analysis ERA5, and the New Europe Wind Atlas (NEWA), for the Gulf of Lion, a region in the Mediterranean Sea with complex coastline (Dimitriadou et al., 2025). The wind direction from the three models was compared to observations from two buoys. Even though ERA5 performed best on wind direction, only for one of the buoys, the SAR-derived wind speed using ERA5 in CMOD5.N resulted in the best statistics compared to 228 collocated buoy winds speed with RMSE 1.40 m/s. For the other buoy, NEWA wind direction input for wind retrieval compared to 77 collocated samples gave the best RMSE of 1.86 m/s.
References:
Badger, M., Fujita, A., Orzel, K., Hatfield, D., Kelly, M. (2023) Wind Retrieval From Constellations of Small SAR Satellites: Potential for Offshore Wind Resource Assessment, Energies 16, no. 9 (2023): 3819, https://doi.org/10.3390/en16093819.
Dimitriadou, K., Olsen, B.T., Badger, M., Hasager, C.B. (2025) SAR offshore wind fields in the Gulf of Lion. J. Appl. Meteorol. Climatol. (accepted)
Hasager, C. B., Imber, J., Fischereit, J., Fujita, A. Dimitriadou, K. & Badger, M. (2024) Wind speed-up in wind farm wakes quantified from satellite SAR and mesoscale modeling. Wind Energy, 2024, 0:1-19. https://doi.org/10.1002/we.2943
Takeyama, Y., & Kurokawa, S. (2024) Development of X-Band Geophysical Model Function for Sea Surface Wind Speed Retrieval with ASNARO-2. Atmosphere, 15(6), 686.
An investigation on the accuracy of SAR wind speeds derived from the Japanese StriX showed comparable results to Sentinel-1 using meteorological observations from a mast in the North Sea. Wind speed from StriX vs. mast data showed RMSE 1.38 m/s and bias -0.42 m/s for 41 collocated samples, while wind speed from Sentinel-1 vs. mast data showed RMSE 1.83 m/s and bias -1.02 m/s for 838 collocated samples (Badger et al., 2023). The geophysical model function (GMF) applied was X-MOD2, developed for the German TerraSAR-X. A new GMF may improve wind speed retrieval.
StriX observed winds over an offshore wind farm in the North Sea, and wake conditions from StriX and Sentinel-1 showed a speed-up effect downwind in 48% and 34% of cases, respectively (Hasager et al., 2024). The result is remarkable, as a common understanding is that the wind speeds are lower downstream of a wind farm due to the wake. At hub-height, wind speeds are lower, but at the 10 m height of SAR wind speed, the mesoscale model applied in the study supported the finding of speed-up (Hasager et al., 2024).
Another small Japanese satellite, ASNARO-2, carries an X-band SAR. It was found that neither X-MOD2 nor an improved version of X-MOD2 suitable for the Italian COSMO-SkyMed was suitable for ASNARO-2. A new GMF for ASNARO-2 was developed using numerical weather model wind speed. The ASNARO-2 wind speed results showed RMSE 1.37 m/s and bias -0.10 m/s for 8128 collocated samples from the weather model, and comparison results RMSE 1.68 m/s and bias 0.03 m/s for 22 collocated samples from ocean buoy data (Takeyama & Kurokawa, 2024).
A major advantage of combining wind speeds from several SAR satellites for wind resource assessment is that the collection will be more frequent and at more times during the day. The latter is particularly relevant for offshore areas with pronounced diurnal wind speed variations, such as land-sea breezes (Badger et al., 2023).
Efforts to improve wind speed retrieval from Sentinel-1 C-band SAR were undertaken using three different established numerical models, namely the Global Forecasting System (GFS), re-analysis ERA5, and the New Europe Wind Atlas (NEWA), for the Gulf of Lion, a region in the Mediterranean Sea with complex coastline (Dimitriadou et al., 2025). The wind direction from the three models was compared to observations from two buoys. Even though ERA5 performed best on wind direction, only for one of the buoys, the SAR-derived wind speed using ERA5 in CMOD5.N resulted in the best statistics compared to 228 collocated buoy winds speed with RMSE 1.40 m/s. For the other buoy, NEWA wind direction input for wind retrieval compared to 77 collocated samples gave the best RMSE of 1.86 m/s.
References:
Badger, M., Fujita, A., Orzel, K., Hatfield, D., Kelly, M. (2023) Wind Retrieval From Constellations of Small SAR Satellites: Potential for Offshore Wind Resource Assessment, Energies 16, no. 9 (2023): 3819, https://doi.org/10.3390/en16093819.
Dimitriadou, K., Olsen, B.T., Badger, M., Hasager, C.B. (2025) SAR offshore wind fields in the Gulf of Lion. J. Appl. Meteorol. Climatol. (accepted)
Hasager, C. B., Imber, J., Fischereit, J., Fujita, A. Dimitriadou, K. & Badger, M. (2024) Wind speed-up in wind farm wakes quantified from satellite SAR and mesoscale modeling. Wind Energy, 2024, 0:1-19. https://doi.org/10.1002/we.2943
Takeyama, Y., & Kurokawa, S. (2024) Development of X-Band Geophysical Model Function for Sea Surface Wind Speed Retrieval with ASNARO-2. Atmosphere, 15(6), 686.