7B.2 Evaluation of the WRF Stochastic Kinetic Energy Backscatter scheme for wind resource assessment

Rife, Daran, Jessica Ma, Rich Whiting, and Melissa Elkinton, GL Garrad Hassan

Mesoscale-model-based virtual time series are increasingly being used as a source of independent, credible reference data for wind resource assessment. The demand for virtual time series is likely to increase as wind farm development expands into regions where few if any high quality long-term reference measurement stations exist. This is especially true at offshore locations, where the cost of conducting a measurement campaign with a single on-site mast is estimated to be in excess of $1M. One of the major limitations of current virtual time series methods used within the wind energy industry is that they are based on deterministic modeling systems. That is, they are created using a single model, with a single configuration, and a single set of inputs, to yield a single answer. These answers include a variety of unavoidable sources of error and uncertainty.

This talk evaluates the stochastic kinetic energy backscatter (SKEB) ensemble technique for creating long-term reference data for wind energy resource assessments. The method is tested at a variety of locations within North America having topographic and wind regime characteristics that range from relatively simple to very complex. We quantify the ability of the SKEB ensemble to reduce uncertainty in the simulated time series relative to deterministic downscaling, and also evaluate its performance relative to the well-established multiphysics ensemble method. An ensemble-calibration technique is then described, and is demonstrated to significantly improve the ensemble mean from both the SKEB and multiphysics methods, and further reduces the uncertainties in the simulated time series.