Parametric and Initial Condition Sensitivity of Wind Ramp Events

This study investigates the sensitivity of wind speed forecasts to variations in parameters within the model physics as well as perturbations to the initial conditions. As part of the second Wind Forecast Improvement Project (WFIP2) wind ramp events in the Columbia River basin are investigated. This work builds on previous studies that indicate a small number of parameters in the surface layer and planetary boundary layer (PBL) schemes are responsible for the majority of the forecast uncertainty. These parameters include the Prandtl number, turbulent length scales, surface roughness, and those related to the dissipation of turbulent kinetic energy. In this study the values of parameters within the Mellor-Yamada-Nakahishi-Niino (MYNN) PBL scheme and the MM5 surface layer scheme of the Weather Research and Forecast (WRF) model are perturbed in a systematic way to evaluate the parametric sensitivity during high impact, wind ramp events. The forecast variance from the physics ensemble is compared to the spread from an initial condition ensemble that uses an ensemble Kalman filter (EnKF) data assimilation framework to evaluate the relative contributions from initial condition and parametric uncertainty.