Building a better wind forecast: A stochastic forecast system using a fully-coupled hydrologic-atmospheric model.

摘要

Wind power is rapidly gaining prominence as a major source of renewable energy. Harnessing this promising energy source is challenging because of the intermittent nature of wind and its propensity to change speed and direction over short time scales. Accurate forecasting tools are critical to support the integration of wind energy into power grids and to maximize its impact on renewable energy portfolios. A numerical weather prediction tool is limited by model errors arising from simplifications in the way it represents the physics of the natural system. Land surface - atmosphere feedbacks are strongly dependent on both atmospheric processes and hydrologic processes at and below the land surface. It has been shown in the literature that improving the physical representation of these feedbacks leads to better forecast results for precipitation distribution and wind speeds. Key to this physical representation is soil moisture distribution. By using PF.WRF, a fully-coupled hydrologic and atmospheric model incorporating the ParFlow hydrologic model in the the Weather Research and Forecasting atmospheric code, it is possible to dynamically simulate water movements in the subsurface generating more realistic soil moisture fields to interact directly with atmospheric processes. This work traces uncertainty propagation from subsurface hydraulic conductivity through soil moisture and latent heat flux and into the atmosphere to analyze its impact on wind speed, the extent of that impact in the presence of prevailing winds, and the length scales over which that impact is important. A data assimilation system using an implementation of the ensemble Kalman filter is developed and verified to reduce uncertainty in simulated wind speed by informing the forecast system with observed soil moisture values, demonstrating that even in a small model domain wind speed is sensitive to variation in soil moisture distribution.