Utilization of High-Resolution Weather Modeling to Improve Downsloping Wind Event Prediction for Vermont Utility Applications

Vermont's complex terrain and vast local variability coupled with large gaps in observational data pose many challenges from a weather forecasting perspective. In particular, high wind events via downsloping are a common occurrence throughout much of the state and are among the most frequent causes for power outages. The increasing trend of severe weather phenomena over the past few years has put electric utilities under high risk. Thus, improving our ability to understand and forecast high wind events is critical not just to ensure a reliable electric grid, but to our very ability to maintain a safe and functioning society. In a collaborative effort to increase the resiliency of Vermont's electrical grid, Vermont Electric Power Company (VELCO) and statewide partners developed the Vermont Weather Analytics Center (VTWAC) to increase grid reliability, lower weather event-related operational costs, and optimize the utilization of renewable generation resources.

The weather prediction component is powered by IBM's Deep Thunder, an advanced NWP model that is based, in part, on a configuration of the Advanced Research core of the Weather Research and Forecasting (WRF-ARW) model. Deep Thunder runs two 72-hour forecasts daily at 1 km horizontal resolution and outputs variables at 10 minute intervals. Vertical resolution is also high with 51 vertical levels in order to account for characteristics of the wind turbines. Deep Thunder uses RAP for background fields and NAM for lateral boundary conditions, as well as complex physics configurations to account for highly rural and urban environments. The current wind gust algorithm utilizes a TKE-based PBL scheme which derives wind gusts using a TKE/buoyancy algorithm. Since transitioning from 2 km to 1 km resolution, the operational model has shown significant improvements in resolving surface wind gusts which has aided utility impact forecasts.

The presentation will first provide the motivation for the VTWAC project which will include a discussion on the synoptic and mesoscale forecasting challenges within Vermont. Model specifications will then be shared which will highlight the resolution differences between the operational 1 km and previous 2 km versions. Qualitative and quantitative verification of a few impactful events will then be presented. Potential future work and applications of the operational weather model will also be discussed which include outage/impact prediction, road weather forecasting, recreational forecasting, and climate change.