Near-shore wind resource estimation using lidar measurements and modelling
Rogier Floors, Andrea N. Hahmann, Alfredo Peņa
DTU Wind Energy, Roskilde, Denmark
There is a worldwide interest in reducing the costs of offshore wind energy, which can be achieved by cheaper and more accurate estimations of the wind resources before wind farms are installed. The RUNE project aimed to reduce the uncertainty in near-shore wind resources using long-range scanning lidars and meso- and microscale modelling. Onshore scanning lidars were measuring winds on a transect covering positions over water up to 5 km and over land up to 4 km across the coast Danish west coast.
The mesoscale WRF and a microscale linearized model were used to predict the complex coastal flow. The WRF model was run during the same period as the measurement campaign, and the sensitivity to its physical set-up options were investigated. Over water, the WRF model wind speed outputs agreed well with predictions of the wind speed using the observations together with the microscale model. Over land, however, there were larger differences due to microscale effects and a speed-up near the shore.
The RUNE experiment took place during four months starting in November 2015. Three scanning lidars were positioned at the coast and measured the wind up to 5 km offshore and 4 km inland. Four additional vertically profiling lidars were installed and a floating buoy measured the wind offshore. The WRF model was set-up and run in 16 different configurations to find the best configuration for predicting the near-shore wind speed. The sensitivity to usage of different boundary-layer schemes, different land-cover descriptions, different sea-surface temperature descriptions, different atmospheric boundary conditions and different horizontal grid spacings were evaluated. Furthermore, the linearized flow model that is embedded in WAsP was used to predict the offshore mean wind speed from onshore profiling lidar measurements.
The scanning lidar measurements compared well with vertically profiling lidars at overlapping positions. When the scanning lidar measurements were available, the WRF model mean wind speed agreed within 3% with the measurements. The WRF simulations using the MYJ boundary-layer scheme, CORINE land cover description and ERA-interim boundary conditions had a lower root-mean-square error than simulations using the YSU scheme, the USGS land description and FNL boundary conditions, respectively. Over land, none of the schemes represented the microscale effects well and the linearized flow model driven by land measurements was closer to the observations. Offshore, the WAsP-extrapolated observations and the WRF estimated mean wind speed generally agreed within 2-3%.
Scanning lidar measurements and offshore wind modelling can potentially reduce the cost of offshore wind estimations, because offshore measurement campaigns are expensive and hard to perform. When comparing the scanning lidar measurements with vertical profiling lidars, agreement is very good (correlation coefficient near 1.0). However, the availability of the lidar measurements at distances of several kilometersis an issue that has to be addressed. The mesoscale model simulations compared very well with the measurements offshore, but failed to represent the flow near the coast and inland, even in such simple terrain.
After this presentation the audience will have an increased understanding of wind resource estimations in the coastal zone, how the flow can be modelled with both micro- and mesoscale models and what are the pitfalls both when dealing with measurements and modelling. The possibility of estimating the wind offshore with scanning lidars is discussed.