Overview

The Planetary Boundary Layer (PBL) wind is one of most difficult weather variables to predict because it is a result of multiple scale interactions and it typically varies dramatically in space and time. This is particularly true when complex terrain is present. Although modern mesoscale numerical models contain sophisticated PBL parameterizations, such as non-local mixing schemes and Turbulent Kinetic Energy (TKE) - based methods, accurately forecasting PBL winds is very challenging. Wind-energy forecasting critically relies on accurate prediction of the wind in the PBL. Our research goal is to study the PBL processes, and especially the wind characteristics in the lowest 200–300 m, at wind farms, and to acquire fundamental knowledge for developing an innovative wind-energy forecasting capability.

Description

To study the impact of PBL processes on wind-energy production, we focused on a large wind farm located in northern Colorado, the Cedar Creek wind farm. There are two high meteorological-towers and 274 wind turbines in the Cedar Creek farm, spreading over a hilly area with dimensions of about 10 km by 15 km. A field campaign with intensive measurements is being jointly carried out at the farm by Xcel Energy, Vaisala, and NCAR. Analysis of the data shows large wind variations across the farm, with the wind patterns changing rapidly with time. The farm-wide wind-speed variance varies greatly with weather regime.  

The NCAR WRF-based RTFDDA system has been adapted for multiscale simulation of the weather at the Cedar Creek farm, encompassing both mesoscale to microscale processes. The model has been set up to simulate a weather event during 14–16 November 2008. Six nested domains were established, with grid increments of 30, 10, 3.3, 1.1, 0.370 and 0.123 km. PBL parameterizations were applied for the four coarse meshes, whereas the two finest meshes were run with the LES settings. The WRF “observation-nudging”-based FDDA was activated on the four coarse meshes, which continuously assimilated diverse synoptic and asynoptic weather observations, and provided accurate lateral-boundary forcing to the high-resolution LES modeling.  

The modeling results show that the WRF-based RTFDDA LES model reproduces many observed intra-farm flow features. The study also illustrates the existence of large differences in the simulated wind shears and speeds in the PBL for different model grid sizes.

Comparison of WRF-RTFDDA-LES modeled and observed winds at the turbine hub-heights in the Cedar Creek wind farm in northern Colorado.