SITING A WIND PROJECT - Wind Resource Assessment: A Practical Guide to Developing a Wind Project - page 26

• surrounding topography, obstacles, and surface roughness
• sensor heights, boom orientations, and distances from tower
• sensor maintenance protocol and records
• duration of data record
• QC and analysis applied to the data.

Wind data tend to be more representative of the surrounding area where the terrain
is relatively flat. In complex terrain or near coastlines, the ability to reliably extrapolate
the information beyond a station's immediate vicinity is more limited and may require
expert judgment and wind flow modeling. Even in flat terrain, good exposure to the
wind is essential, especially for short towers. Measurements taken in obstructed areas
or on rooftops should not be used unless there is good reason to believe that the effects
of the obstructions are small. In any event, data from existing meteorological towers,
except in rare instances, are unlikely to be able to replace on-site measurements from
a wind monitoring campaign.
These days, towers instrumented for wind resource assessment are typically at least
50 m tall, whereas towers installed in the 1980s and 1990s, for which data may be
available, were often only 20-30 m tall. The heights of most general-purpose meteo-
rological towers range from 3 to 20 m, and the international standard is 10 m. When
comparing data from different stations, mean wind speeds should be extrapolated to
a common reference height (e.g., 80 m, a typical wind turbine hub height). This can
be done conveniently using the power law:

v 1 h 2
h 1

α

v 2 = (3.1)

where

v 2 = the projected speed at the desired height h 2

v 1 = the observed speed at the measurement height h 1

α =

a nondimensional wind shear exponent.

The uncertainty in the projected speed depends on both the ratio of heights and the
uncertainty in the wind shear exponent, as can be seen in the following equation
(derived from Eq. 3.1):

ln h 2
h 1

σ v = 100 σ α ( % ) (3.2)

is the uncertainty 1 Here, σ v the uncertainty in the

projected speed in percent, and ln is the natural logarithm. For example, suppose

σ α in the shear exponent and

1 Here and elsewhere in this topic, the uncertainty is defined as the standard error , σ , which is the stan-
dard deviation of the distribution of possible results of a measurement. For normally distributed data, a
measurement will fall within one standard error of the mean of the distribution about 68% of the time.