Chemistry Reference
In-Depth Information
period described here) was the most important component of the heat flux,
forcing a cool skin. Relative humidity varied in the range 69-77 %, adding
a latent heat flux to the net heat loss. Air and water temperatures were
similar, and consequently the sensible heat flux was quite small. The ther-
mal noise of each pixel (= 80 mK) obscures all but the strongest features
At moderate wind speeds, by far the greatest signal is associated with the
crests of actively breaking waves, but these were fairly rare. It is also ap-
parent that the wakes of these breaking waves are relatively warm. An im-
age of a relatively large breaking wave is shown below (Figure 1).
The original images do not usually show any distinct features apart from
an occasional breaking wave. In order to extract more information, we
have processed the images as follows. The image is warped (with the
greatest possible preservation of individual pixels) to a 128 x 128 pixel,
6.4 m x 6.4 m subsample at the centre of the original image. Warping is
necessary, but the mapping was carefully designed to remain as close as
possible to mapping one original pixel to a single final pixel. Linear hori-
zontal gradients are eliminated (de-trended images). Very large scale (>3
m) variability which can not be sampled satisfactorily is removed. The to-
tal standard deviation within these images is dominated by the 80 mK
“white noise” of the instrument (a natural limit of the sensing element at its
normal operating temperature). The variation with time of the standard de-
viation of temperature within each image is shown in Figure 2 (upper left).
The remaining signal is highly variable. Dramatic increases and then falls
in standard deviation are observed on time scales of 10 minutes and less. A
few isolated values can be attributed to special features; for example the
breaking wave featured in Figure 1 is responsible for a single high value at
~2151 UTC. The slower variation (gradual increase followed by decrease)
follows the trend in heat flux, but the excursions are unexpected. The
small-scale (0.1-1 metre) fluctuations slightly exceed the instrumental
noise, but generally any coherent fluctuations at these scales are not suffi-
ciently large to extract from the white noise. The standard deviation within
images is separated according to spectral range. The variability in two
spectral ranges is highly correlated over time periods of a few minutes
(e.g., Figure 2, lower left), but the two sets of values are not simply pro-
portional (straight line in figure) as might be expected if each was propor-
tional to the instantaneous heat flux or reflectivity. Larger scale variability
contributes disproportionately to the rapid excursions in standard devia-
tion.
