222 
PACIFIC SCIENCE, Vol. Ill, July, 1949 
If, on the other hand, it can be shown that 
for certain specific purposes a reduction in 
average error of estimate outweighs the im- 
portance of very large errors in estimate for 
individual years, application of the technique 
should be considered. 
The work of Solot ( 1948) has opened up new 
fields for the growth of longer-range forecasts 
for Hawaii. Having classified individual months 
on the basis of average rainfall for the Territory 
of Hawaii and on the average number of days 
of rain, he segregated monthly mean surface 
pressure maps for the Pacific area. The anoma- 
lies of mean pressure for individual months from 
the long-time average pattern were distinctly 
different for wet and dry months. Wet months 
are characterized by higher than normal pres- 
sure in the Aleutian area and a weak low in the 
mean pressure field over Hawaii. Dry months, 
on the other hand, have abnormally low pres- 
sure near the Aleutians and higher than normal 
pressure over the Hawaiian Islands. 
Because there is some month-to-month per- 
sistency in the patterns most different from 
normal, these pressure anomaly patterns have 
some forecast value. Solot found that near- 
normal pressure distributions apparently have 
no significant persistence and the forecast value 
is dubious. 
This distribution of anomalies fits the general 
knowledge of Hawaiian rainfall. Wet months, at 
least in winter, are characterized by a dislocation 
of the Pacific high and the admittance of 
cyclones to more southerly paths. 
Riehl (MS.) interprets the import of Solot’s 
correlations of pressure patterns and rainfall 
as follows: Mean monthly pressure patterns do 
not indicate deviations from average conditions 
which "would tend to raise or suppress rain- 
fall throughout the month. They denote a basic 
state of the general circulation favorable or un- 
favorable for generation of a few potent dis- 
turbances." 
Another possible forecast tool is provided by 
the work of Yeh (MS.). He found that there 
is some general relation between the latitude 
of the belt of maximum west wind at 10,000 
feet (the jet stream of Rossby) and precipita- 
tion in Hawaii. When the jet stream lies south 
of 40° N. Lat. on a given day, the probability 
of rainfall over Hawaii is small. Positions be- 
tween 50° and 60° N. apparently are associated 
with higher rainfall probability. 
WINDS 
Most of the older studies of upper air data 
over the Pacific are of little significance in the 
development of forecast techniques. Beals 
( 1921b ) presented wind roses for various levels 
above Pearl Harbor, later extended to higher 
levels by Thomson (1928). The latter showed 
that the west winds, which surmount the low- 
level trade wind easterlies, reach lowest eleva- 
tions in March and April on the average, while 
the height of the easterlies is greatest in May 
and June. 
Beals ( 1927*0 showed that the surface wind 
at Honolulu progressively changed in mean 
direction from northeasterly to a more easterly 
flow between 1905 and 1924. Wentworth 
(1949) has brought this study up to date and 
finds that in the last decade the earlier trend has 
been reversed, the mean winds gradually backing 
to a more northeasterly direction. 
The diurnal patterns of wind have been dis- 
cussed by Beals ( 1921a), Henry (1925), and 
Leopold (1948). In summary, sea and land 
breeze regimes control the wind direction in 
the lowest levels on the protected coasts of all 
the islands. 
Leopold (1948) noted that the line along 
which the sea breeze meets the trade wind is 
associated with the development of a cloud line 
on Lanai, Molokai, and northeastern Hawaii. 
Diurnal changes in cloudiness in many areas are 
related to the development of sea and land 
winds. The regime is particularly apparent along 
the Hamakua (northeast) coast of Hawaii 
where drainage winds from Mauna Kea keep 
the slope clear at night; during the day the sea 
breeze reinforces the trades and causes an after- 
noon maximum of cloudiness. 
