The sampling variability and the validation of high frequency radar measurements of the sea surface.

Remote sensing is becoming an increasingly important tool for ocean wave
measurement, and over the past decade much progress has been made in
the development of the wave measuring capabilities of HF (High Frequency)
radar. This system is able to make detailed and near continuous observations
of the sea surface over a wide area. However, because the mathematics of
the data extraction process is rather difficult, the statistical properties of the
observed data have to date been poorly understood.
In this study, the approximate sampling distributions of a variety of measurements
from HF radar (including significant waveheight, mean wave period,
wind direction, and various spectral parameters) are derived in terms
of quantities that are either known or estimable. The resulting confidence
intervals are, in the case of significant waveheight and mean wave period,
of comparable width to those obtained from the corresponding NURWEC2
(Netherlands UK Radar Wave buoy Experimental Comparison) wave buoy
measurements, and in the case of spectral power, they are narrower.
Furthermore, methods are derived by which such radar measurements may be
compared with their corresponding wave buoy measurements in a statistically
valid manner, and their relative biases estimated. These methods are then
applied to data taken during the NURWEC2 field trial, which suggest that
the radars and the wave buoy show good correspondence for measurements
of significant waveheight and of spectral power (over 85
-
125mHz
-
the
frequencies with most wave power, and hence those of most importance).
There is also a fair correspondence for mean period measurements in the
range 6.8
-
11.0secs. Spectral mean direction shows good correspondence
over 85
-
155mHz over the somewhat limited directional range (i. e. as
observed during the NURWEC2 storm) of the data.