Numerical weather prediction (NWP) models are widely used in Wind Energy operations as a base models for the wind forecast. Continuous updates and improvement of these models with a better physics, parameterization schemes, and the horizontal grid resolution require accurate assessment of model performance in various landscapes and atmospheric conditions. In this study, the performance of two models developed at the ESRL of the NOAA investigated by measurements from scanning Doppler lidars during the Wind Forecast improvement Project-2 (WFIP-2) in the complex terrain of the Columbia River Valley. This 18-month long project, sponsored by DOE and NOAA, is aiming to improve forecasting of wind flow complicated by mountainous terrain, coastal effects, and the presence of numerous wind farms in this area. The accuracy of the lidar measurements, assessed during previous field campaigns and post-processing data from WFIP2, was considered to be sufficient to investigate the ability of NWP models forecast challenging wind flow conditions in complex terrain.
The paper presents results on the model validation by Doppler lidar measurements at three research sites, the assessment of model skill to capture spatial and temporal variability of wind-flow profiles and quantifies model errors to forecast wind speed at the height of 80 m AGL, the hub-height of the most wind turbine in the surrounding area, and through the layer occupied by turbine blades. All models captured general trends of diurnal wind flow variability but show some temporal and vertical discrepancies with larger biases between measured and modeled wind speed in the first 500 m AGL. The biases were dominated by errors during frequently occurring complex-terrain flow systems—cold pools in winter months and westerly gap flows in spring-summer months. These error characteristics were also identified in the behavior of the seasonal and annual validation statistics, computed for each site and for a three sites composites.
Knowledge of the error in the hub-height wind speed forecasts is of critical economic importance for the calculation of energy produced by wind turbine.