Verification of Hub-Height Wind Forecasts from the NCAR-Xcel Ensemble-RTFDDA System

In support of Xcel Energy wind power prediction, NCAR has implemented an operational E-RTFDDA system (Ensemble Real-Time Four Dimensional Data Assimilation and forecasting system) over the western-central states. The ensemble system runs with two nested domains at 30 and 10 km grid intervals, and its fine-mesh 10-km domain covers the Rocky Mountains from New Mexico to Montana, and also the western Midwestern States, and most parts of the Central/South Plains. E-RTFDDA is a WRF and MM5 based multi-model, multi-perturbation approach ensemble system. It is able to produce ensemble 4D continuous data assimilation and forecasts. This system has been running operationally since May 2010, with 6-hour forecast cycles, providing forecasts at up to 48 hours in each cycle. In this study, the operational E-RTFDDA hub-height wind forecasts during the year 2011 are verified against the wind observations at selected wind farms located in Minnesota, Colorado and Texas, which features dramatically different weather regimes. The operational model outputs are analyzed to assess overall ensemble performance and properties, with focus on the systematic errors that can vary significantly with and within the wind farms. The research in this paper is also focused on the ensemble forecast performances in terms of the “wind-ramp” probabilistic forecast using the raw ensemble model output, and the “climatological” characteristics of several WRF physics settings. This study also evaluates the additional skills from an ensemble calibration process based on an analog-based bias-correction and a quantile-regression algorithm, that are part of the operational system.