Regime-based forecast performance during WFIP 1

Abstract:

The principal objectives of the first Wind Forecast Improvement Project (WFIP 1) were to improve short-term (0 - 6 hr) wind power forecasts through the assimilation of targeted remote sensing and surface observations with an enhanced model ensemble forcast system. The WFIP 1 field deployment/modeling campaign in the Southern Study Area (SSA--encompassing most of central and western Texas) ran from August 2011 through Septembe 2012. This ensured observational data and model output for all representative weather regimes affecting the SSA. Cold and warm season regimes featured synoptic-scale, convective, and low-level jet (LLJ) phenomena that are responsible for the favorable wind resource in the SSA, and also posed a challenge for assigning specific explanations for the observed forecast improvements (e.g. additional observations, model improvements, or a combination of both). LLJs produced hourly capacity factors exceeding 80% in aggregated wind farm power production, while synoptic-scale systems were responsible for the largest ramp events observed during WFIP 1. Accurately forecasting convective phenomena (such as outflow boundaries) during WFIP 1 was at times problematic. Here, we present regime-based and phenomenological-related forecast performance results for WFIP 1. These performance metrics suggest future research pathways that will facilitate improvements in operational wind power forecasts.