Use of synthetic aperture radar for offshore wind resource assessment and wind farm development in the UK
Cameron, Iain Dickson
The UK has an abundant offshore wind resource with offshore wind farming set to grow rapidly over the coming years. Optimisation of energy production is of the utmost importance and accurate estimates of wind speed distributions are critical for the planning process. Synthetic aperture radar (SAR) data can provide synoptic, wide area wind field estimates at resolutions of a few kilometres and has great potential for wind resource assessment. This thesis addresses the key challenges for the operational implementation of SAR in this context; namely the accuracy of SAR wind retrievals and the ability of SAR to characterise the mean wind speed and wind power density. We consider the main stages of SAR wind retrieval; the retrieval algorithm; sources of a priori information; the optimal configuration of the retrieval system; and the challenges for and accuracy of SAR wind resource estimation. This study was conducted for the eastern Irish Sea in the UK, a region undergoing significant offshore wind energy development. A new wind retrieval algorithm was developed that implements a maximum a posterior probability (MAP) method drawn from Bayesian statistics. MAP was demonstrated to be less sensitive to input errors than the standard direction-based wind speed algorithm (DWSA) and provides a simple retrieval quality check via the error reduction ratio. Retrieval accuracy is strongly influenced by the quality of a priori information. The accuracy of two operationally viable a priori sources, mesoscale numerical weather prediction (NWP) data and WISAR image directions, was evaluated by comparison against in-situ wind observations and WERA coastal data. Results show that NWP wind speeds produce good wind speed and direction estimates with standard deviations of ¬±2 ms-1 and ±16o respectively. WISAR directions were less accurate producing standard deviations ranging from ±20o to ±29o, but were preferable when strong differences between NWP timesteps were observed. The accuracy of SAR wind retrievals was evaluated by comparison against in-situ wind observations. The MAP algorithm was found to provide modest improvements in retrieval accuracy over DWSA. Highest quality retrievals achieved using the CMOD5 forward model, producing wind speeds with a RMSE of 1.83 ms-1. Regarding the ability of SAR to estimate offshore wind resources, dataset density was found to be a controlling parameter. With 103 scenes available mean wind speeds were well characterised by comparison against in-situ observations and Wind Atlas results, while wind power density showed considerable errors. The accuracy of wind speed maps was further improved by accounting for wind direction and fetch effects upon the SAR wind distribution. A key strength of the SAR wind fields is their ability to identify the effect of mesoscale structures upon the surface wind field with atmospheric gravity waves observed in 30% of the images. These structures are shown to introduce wind speed fluctuations of up to ±2 ms-1 at scales of 5 to 10 km and may have significant implications for wind power prediction. These findings show that SAR may provide an important source of wide area wind speed observations as a complement to existing wind resource estimation techniques. SAR may be of particular use in coastal areas where complex wind fields are observed.