UNCERTAINTY IN WIND RESOURCE ASSESSMENT - Wind Resource Assessment: A Practical Guide to Developing a Wind Project - page 224

If, on the other hand, the uncertainties are perfectly correlated, then the combined
uncertainty is their linear sum, 0 4%, which is 40% larger.

The challenge is to determine which errors are correlated and which are not. (Some
errors may be partially correlated; however, such cases can usually be dealt with by
dividing the uncertainties into separate components that are either purely correlated
or purely uncorrelated.) By convention, the different components of the uncertainty
for a single mast are assumed to be uncorrelated and can therefore be combined as
the sum of the squares. Although one can think of situations where this does not hold
completely true, 6 it is generally a good assumption. The uncertainty ranges discussed
earlier for the separate components can therefore be combined into an overall range
for a single mast, as shown in Table 15-1.
When uncertainty estimates from different masts are combined, more attention must
be paid to the possibility of correlated errors. Here are some examples:

. 02 + 0 . 02 =

• When the masts at a project site span the same or a similar period, their long-term
mean wind speed estimates may be biased in a similar way.
• Any deviation of the future wind resource from the historical average will apply
equally to all parts of the project, regardless of the number of masts used in the
resource assessment.
• Wind flow modeling errors can be systematically biased if the masts are situated
in one type of terrain (for example, along a ridgeline), while some of the turbines
are in a different type of terrain (down the slope).
• Tower effects on speed measurements may introduce a consistent bias if the
anemometers are all mounted in the same direction relative to the prevailing
wind (of particular concern when there is only one anemometer mounted at each
height).

Table 15-1. Summary of general uncertainty ranges by category for a single mast,
equipped, maintained, and analyzed according to the guidelines described in this
topic. For an explanation of each uncertainty range, see the corresponding text.

Category Uncertainty, %

Measurement accuracy (single anemometer) 1.0-2.5
Historical wind resource 1.6-4.0
Future wind resource (plant life of 10 or 25 yr) 1.4-2.2
Wind shear 0.0-6.3
Wind flow modeling 2.0-10.0
Total uncertainty 3 - 13
It is assumed that the various components are uncorrelated with one another, and thus the total uncertainty
is the square root of the sum of the squares of the individual uncertainties.
Source : AWS Truepower.

6 One such case was mentioned earlier: turbulence or vertical winds may bias all anemometers in a similar
way. When this happens, averaging the wind speeds from two anemometers at the same height does not
reduce the uncertainty. As another example, when the uppermost anemometers are placed too close to the
tower top, there may be a tendency to overestimate both the mean speed and the shear.