Comparison and validation of high-resolution techniques for spatial mapping of wind energy resources

An important part of the process of developing wind energy projects is the assessment of the wind resource across the specific project site. Assessment begins with an initial search for particularly windy areas at a coarse spatial scale (the “prospecting” phase), and then focuses in to the specific area of the proposed wind farm to determine the most productive layout of individual wind turbines. Terrain complexity increases the challenge of making accurate wind energy assessments at fine spatial scales. A variety of tools for assessing potential wind energy production are used at each stage of the project development, depending on the accuracy demanded for that stage. Many of these tools include models of varying sophistication which, when combined with on-site observations via statistical techniques, can be used to spread observed information both spatially, so that limited observations can be used to produce detailed spatial maps of wind resource, and temporally, so that short-term observations can be placed in a long-term context. Often more than one model or method are employed to simulate processes at different time and spatial scales.

3TIER, Inc., a renewable energy information services company, had a unique opportunity to test a variety of modeling techniques to assess the wind resource in a region of moderately complex terrain in the western United States, where multiple meteorological towers were deployed. Different towers and time periods were used to train the techniques and test them. The methods tested included:

* WRF alone, using different horizontal resolutions.

* WRF at coarser horizontal resolution coupled with a high-resolution downscaling model that makes boundary-layer flow adjustments based on high-resolution topography and land-use data.

* A linear flow model (Wind Atlas Analysis and Application Program, or WAsP), applied to areas surrounding the individual tower observations.

The tested techniques represent not only different approaches and representations of atmospheric physics, but also widely different computational costs. Therefore, cost-benefit is an important focus of the comparative analysis. The performance of the methods will be assessed based on standard metrics such as bias and root mean-squared error of wind speed and wind power. The performance of the different methods will be discussed, with emphasis on physical interpretation, as well as the cost-benefit perspective.