Wind Resource Assessment Utilizing Time-Averaged Community Earth System Model data

With the availability of improved computational capabilities, global and regional climate model simulations are being carried out at increasingly higher spatial resolutions as part of climate assessment studies. The finer spatial grid of these simulations provides greater accuracy for a variety of applications such as estimating hurricane trends, climate extremes, and wind resource assessment.

Although spatial resolution has increased, often only time-averaged information, such as the monthly mean, is saved for many climate model simulations to minimize the amount of the archived data. As wind resource assessment requires information with high temporal resolution, statistical corrections such as energy pattern factors have to be calculated in order to properly assess accurate wind power density totals.

This study investigates the use of time-averaged climate model data from the Community Earth System Model (CESM) to assess the wind speed/power variability at the typical turbine hub height of 80 meters. The CESM integrations were carried out at a fairly high spatial resolution of 1/4 degree, but they do not have (like most climate models) a level of data at 80 meters. Therefore, extrapolation of surface wind speeds at 10 meters is required.

In order to increase the accuracy of wind resource assessment utilizing data from this model, two corrections were made. The first involved interpolating the wind speed at 10 meters and the lowest pressure level of the CESM to 80 meters instead of using the normal power law exponent of 1/7. The second correction involved computation of the best-fit Weibull distribution at each grid point to estimate the energy pattern factor that is used to correct for the lack of temporal resolution. Before using this technique on CESM data, it was first verified using high temporal resolution data from the North American Regional Reanalysis (NARR) dataset. The wind resource assessment from the CESM integrations is then compared to the NARR dataset, averaged over the period 2003-2012, to assess the fidelity of the climate model simulations.