Wind resource estimation onshore remains a large source of uncertainty for wind energy projects, especially over complex terrain. Over this type of landscape, we observe low-level jets, separated flow, hydraulic jumps and internal gravity waves. In the wind energy relevant, double-ridge terrain near Perdigão in Portugal presented here recirculation appears very frequently and that internal gravity waves are present in the stable flow every other night. These phenomena may affect the wind conditions significantly. For example, the stronger the recirculation behind the upstream ridge, the lower the wind but it is uncertain to what extent. They are usually not included in models used for wind resources assessment.
The campaign at Perdigão had participation from a number of American and European institutions making it one of the largest experiments of its kind. It is a part of the New European Wind Atlas (NEWA) project. More than fifty meteorological masts and twenty advanced scanning wind lidars together with a large selection of other in situ and remote sensing instruments mapped the wind conditions over the terrain.
The measured wind time series and their statistics were compared to a dynamical downscaling methodology based on a model-chain approach. The method is based on forcing, through one-way coupling, a simplified atmospheric unsteady-RANS microscale flow model (EllipSys3D-ABL) by mesoscale dynamical tendencies and surface temperatures generated from the Weather Research and Forecasting (WRF) model. The time- and height-varying tendencies are added directly as source terms into the governing momentum and temperature equations of the microscale model. The tendencies account for the large-scale variations from scales beyond the geometry of the microscale grid.
The simulations centered at Perdigão were made using a horizontal grid spacing of 80 m in the microscale model grid, with forcing extracted from a WRF simulation using 9 km grid spacing. The simulations covered several spring/summer weeks, allowing for a detailed study of specific flow phenomena, such as the occurrence of low altitude jets above the hills, which was captured by the multi-lidar scans at the site. Furthermore it allowed for analysis of the statistical wind speed characteristics of the model results, which are compared to point-measurements from nine tall meteorological masts at the site during the same period. The results shows that the high-resolution microscale model, forced by the coarse mesoscale model tendencies, was able to capture many of the local flow effects at the site, resulting in improvements in the statistical characteristics of the simulated flow compared to the a reference WRF simulation, downscaled to 1 km grid spacing. Overall the results are encouraging and could be used as the dynamical model-chain-component in a future statistical-dynamical downscaling methodology for estimating wind resources in complex terrain.