Assessing Boundary Condition and Parametric Uncertainty in Numerical-Weather-Prediction-Modeled, Long-Term Offshore Wind Speed Through Machine Learning and Analog Ensemble

Abstract

Abstract. To accurately plan and manage wind power plants, not only does the time-varying wind resource at the site of interest need to be assessed, but also the uncertainty connected to this estimate. Numerical weather prediction (NWP) models at the mesoscale represent a valuable way to characterize the wind resource offshore, given the challenges connected with measuring hub height wind speed. The boundary condition and parametric uncertainty associated with modeled wind speed is often estimated by running a model ensemble. However, creating an NWP ensemble of long-term wind resource data over a large region represents a computational challenge. Here, we propose two approaches to temporally extrapolate wind speed boundary condition and parametric uncertainty using a more convenient setup where a mesoscale ensemble is run over a short-term period (1 year), and only a single model covers the desired long-term period (20 year). We quantify hub-height wind speed boundary condition and parametric uncertainty from the short-term model ensemble as its normalized across-ensemble standard deviation. Then, we develop and apply a gradient-boosting model and an analog ensemble approach to temporally extrapolate such uncertainty to the full 20-year period, where only a single model run is available. As a test case, we consider offshore wind resource characterization in the California Outer Continental Shelf. Both the proposed approaches provide accurate estimates of the long-term wind speed boundary condition and parametric uncertainty across the region (R2 > 0.75), with the gradient-boosting model slightly outperforming the analog ensemble in terms of bias and centered root-mean-square error. At the three offshore wind energy lease areas in the region, we find a long-term median hourly uncertainty between 10 % and 14 % of the mean hub-height wind speed values. Finally, we assess the physical variability of the uncertainty estimates. In general, we find that the wind speed uncertainty increases closer to land. Also, neutral conditions have smaller uncertainty than the stable and unstable cases, and the modeled wind speed in winter has less boundary condition and parametric sensitivity than summer.

Machine Learning Weather Analogs for Near-Surface Variables

“…AnEn first identifies the M most similar historical forecasts to the current target forecast, and then, the observations corresponding to the selected historical forecasts consist of the ensemble members. The number of analog members depends on the length of the search history: a very small number of analog ensemble members could lead to sampling errors and a very large number of members could generate extremely unrepresentative ensembles (Delle Monache et al 2013;Hu et al 2021;Bodini et al 2021). This process is repeated for each forecast cycle time (e.g., when the forecast was initiated), each forecast lead time, and at each grid point independently.…”

Machine Learning Weather Analogs for Renewable Forecasting

“…AnEn first identifies the M most similar historical forecasts to the current target forecast and then, the observations corresponding to the selected historical forecasts consist of the ensemble members. The number of analog members depends on the length of the search history: a very small number of analog ensemble members could lead to sampling errors and a very large number of members could generate extremely unrepresentative ensembles [1,22,23]. This process is repeated for each forecast cycle time (e.g., when the forecast was initiated), each forecast lead time, and at each grid point independently.…”

The sensitivity of the Fitch wind farm parameterization to a three-dimensional planetary boundary layer scheme

“…They found that the optimal PBL scheme varies with stability: at this site, MYJ (Janjić, 1994) performed best under stable conditions, ACM2 (Pleim, 2007) performed best under neutral conditions, and YSU (Hong et al, 2006) performed best under unstable conditions. Wind atlases that characterize model uncertainty often employ ensembles of simulations where model inputs, such as PBL scheme, are varied (Bodini et al, 2021a).…”