Figure 9. Grey scale 3-day composite images of waveheight in the north-east Pacific 
showing calm areas dark and rough areas light. Height contours of half meter 
intervals in H] /3 are shown as heavy lines. Left, Figure 9a: composite image for 
November 14-16, 1987. Right, Figure 9b: composite image for November 17-19, 
1987. 
The quasi-random distribution of sea surface heights limits 
the precision of all waveheight measurements. Increased 
precision is only achievable by increased averaging, either in 
space or in time. The standard 20 minutes of single point 
data from a wave-rider buoy is equivalent to about 100 wave 
periods, of 10 second, 156m waves, giving about a 10% 
accuracy in waveheight. For the satellite, the shape of a 
single radar altimeter pulse is affected by the speckle noise 
in the coherent microwave data. Geosat data is averaged 
over 1 second, during which time the altimeter transmits 
and receives 1000 pulses. The average pulse shape should 
therefore be stable to about 3%. 
The leading-edge shape of an individual pulse return is 
determined by the average wave field over a circle centered 
on the nadir point directly beneath the satellite. The radius 
of the circle increases from about 2km for a calm sea to 8km 
for waves with H| /3 = 10 meters. If sea surface heights are 
considered to be conerent over an area equal to about the 
square of the peak energy wavelength, then for 156m waves 
there would be about 2000 independent samples over a 4km 
radius circle, giving over an order of magnitude more 
averaging than from a wave-rider. In one second the 
satellite nadir point will have moved about 6.5km, so that 
the next 1 second average covers a nearly independent area. 
In their amount of averaging, the waveheight measurements 
from a satellite altimeter are superior to those collected by 
surface buoy measurements. However, there is no exact and 
accepted standard against which both techniques can be 
compared. Certainly satellite results show high self- 
consistency, and agreement with the buoy results is within 
satellite specifications (Dobson et al, 1987, Monaldo, 1988). 
The nature of the ocean wave/radar interaction may 
introduce systematic errors in the satellite measurements. 
The sampling areas of the altimeter waveheight 
measurements are very small compared to the distance 
between tracks on a given day. Effectively, the 
measurements are made only along the satellite’s track, and 
the coverage from a single satellite severely limits the 
accuracy of maps compiled by interpolating between the 
tracks. 
A much more frequent coverage than the 17 day repeat 
cycle is needed to avoid aliasing of varying or moving storms 
over the repeat period. In a single day, satellite tracks are 
spaced about 3000km apart at the equator, or 2000km at 
45°N. In 3 days the Geosat orbit has an approximate repeat 
pattern in which these separations are reduced to about 
1000km (700km at 45°N). Ascending and descending passes 
combine to diagonal grids of tracks of this spacing. 
Three days is longer than optimum, and 1000km is also too 
large in the spatial domain, so the coverage from a single 
satellite is again inadequate. However, for the examples that 
follow, the 3 day pattern is used as the best compromise. 
Figure 9 shows two waveheight maps of the north-east 
Pacific compiled from consecutive 3 day sequences of data. 
Positions of the satellite tracks from which data are used, 
are shown. Data is interpolated to fill the entire area using a 
gaussian weighting of data with distance as in Figures 3 and 
4, but with an e-folding distance of 800km. 
In Figure 9a the ocean is relatively calm, with the highest 
waveheight contour being 3.5m, and the lowest 1.5. In 
Figure 9b an intense storm is centered in the Gulf of Alaska. 
Peak observed waveheights were over 10m, but the 
smoothing of the 3 day composite reduces this to 7.7m. The 
highest contour is 7.5m, the lowest 1.5, as before. 
The shape and data values in the high wave area are 
strongly affected by the positions of the tracks from which 
data is combined. However statistical, wave climate, 
information will not be degraded. Figure 10 shows an image 
of the average waveheight for the period November 8 to 
December 22 1987. This clearly shows the band of high 
waveheights along 45°N (highest contour is 4m) and the 
decrease towards the south and southwest (lowest contour is 
1.5m). Figure 10b shows the standard deviation associated 
with this average. The figure shows two regions where the 
range of measured waveheights is high, one centered at 
47°N and 137°W, close to the storm centre in Figure 9b, 
and the other further west, near 180°W. 
Longer data series are needed to show seasonal and 
interannual variations. For many regions of the world 
Geosat data will provide the most accurate wave climate 
data available. Even in areas well served by conventional 
observations, the consistency, high accuracy and wide spatial 
and temporal coverage from the satellite will give a great 
improvement in quality of data.