A further validation check was performed by
comparing the altimeter global trend values with
numerical model results. Figure S6 (SOM) shows
the trend in mean monthly wind speed calculated
from NCEP (National Centre for Environmental
Protection) global reanalysis results (25). This result is qualitatively consistent with Fig. 1.
The present analysis cannot absolutely deter-
mine why there is a stronger trend in wind speed
than in wave height or the underlying physical
processes responsible for the observed positive
trends in both quantities at extreme values. Ob-
servations of local wave generation (26, 27) in-
dicate that for wind-generated waves, significant
wave height is approximately proportional to wind
speed. Hence, one might initially expect that the
spatial patterns for wind speed and wave height
would be similar. However, the situation is more
complex in that wave height in the open ocean is
a mix of locally generated wind-sea and remotely
generated swell. Although trends in wind-sea
will be correlated with trends in the local wind
speed, it has been shown that swell is influenced
by not only the intensity of generating meteoro-
logical systems but also their frequency (4, 6).
Hence, differing trends in wind-sea (a proxy for
wind speed) and swell are plausible.
wind-sea and swell together with swell propagation direction and a basin-specific analysis. Such
data are not directly available from the altimeter.
The present analysis is aimed at determining whether there is a linear trend over the period of the observations (approximately 23 years).
It does not necessarily follow that the observed
trends are a result of, for instance, global warming. Indeed, interannual-to-decadal variations of
the high-latitude wind belts have been observed,
and Hemer et al. (28) have shown that the wave
climate in the Southern Hemisphere is influenced
by the Southern Annular Mode. A regression analysis between the monthly mean altimeter significant wave height and the Southern Annual Mode
Index showed a weak correlation, with a correlation coefficient up to 0.4 across large areas of the
Southern Ocean. Similar interannual variations
have also been shown to be correlated with wave
heights in the North Atlantic (5, 10, 11, 13, 14).
Hence, it is highly likely that such long-term
oscillations will significantly influence the global ocean wind and wave climate. Because the
present data set is only two decades long, it is
Fig. 2. Color contour plots of the 90th-percentile trend (percent per year). Wind speed is shown at the top and wave height at the bottom. Points that
are statistically significant according to the Seasonal Kendall test are shown with dots.