DATA VALIDATION - Wind Resource Assessment: A Practical Guide to Developing a Wind Project - page 115

and type II (false negative) errors. Sometimes methods designed to minimize one
type of error result in an excessive number of the other type. A good data validation
procedure seeks to minimize both types of error.
In this chapter, techniques appropriate for both QC and data validation are covered
under the validation process.
The validation of remotely sensed data is a more specialized topic, which is outside
the scope of this topic.
9.1 DATA CONVERSION
Depending on the data logger manufacturer and model, the data may first need to be
converted from the logger's raw binary format to a text file, spreadsheet, database, or
some other usable format. Manufacturers of the most widely used data loggers (e.g.,
Campbell Scientific, NRG Systems, Second Wind) provide software to do this, which
is either part of the logger software or runs on a separate computer.
In performing this conversion, the analyst must make sure that settings such as
the wind vane deadband, anemometer transfer function, and time zone are correctly
entered in the conversion software. This may seem like a trivial requirement, but
surprisingly many mistakes occur at this stage. For example, it is not uncommon
for boom orientations and magnetic declinations to be entered incorrectly in the site
documentation or for sensor channel numbers to be switched. These and other common
mistakes, if not caught at the outset, can lead to significant errors in characterizing
the site's wind resource.
For this reason, as a general rule, the analyst should seek independent confirmation
of key information whenever possible. For example, photographs can help confirm
reported sensor heights and boom lengths and orientations, and scatter plots of the
ratios by direction of speeds from paired anemometers can help verify anemometer
boom orientations and designations. If no detailed site documentation is available—or
if the documentation was provided by another party—a visit to the site to obtain or
confirm the required information might be warranted.
Calibrated anemometers should be accompanied by a certificate from the agency
that performed the calibration test. The analyst should check this certificate to confirm
the sensor transfer function and to verify that the sensor test was normal. There
is currently some debate within the wind industry about whether, for calibrated
anemometers, the measured transfer function or an average “consensus” function based
on numerous tests of different anemometers of the same model should be used when
converting raw data. Either method is generally acceptable, although there is evidence
that for NRG #40 and Second Wind C3 cup anemometers, in particular, the consensus
transfer function yields results that tend to match IEC Class I anemometers employed
for power curve testing more closely than the measured functions do (1, 2).
As a matter of good data handling practice, both the raw and converted data should
be preserved in permanent archives. All subsequent data validation and analyses should
be performed on copies of the converted data files. Different file name extensions