Objective Estimates of Hawaiian RainfalP 
Charles K. Stidd^ 
The problem of interpreting mean weather 
maps in terms of rainfall has been studied by 
numerous investigators. Klein (1949), for ex- 
ample, described a method applicable to the 
Tennessee Valley in winter. The Hawaiian 
rainfall problem was studied by Solot (1948 
and 1950), whose methods have been of some 
use in estimating expected Hawaiian rainfall 
from the prognostic maps of the Extended 
Forecast Section of the U. S. Weather Bu- 
reau. While the method, in the hands of a 
practiced user, probably shows as much skill 
as the method to be described in this paper, 
tests (Aubert, MS.) have revealed that Solot’s 
principles may sometimes be interpreted in 
quite different ways by two different fore- 
casters, which fact points to the need for a 
completely objective method for making rain- 
fall estimates. One further reason for a new 
study of the problem at this time is the fact 
that upper-level data were not readily avail- 
able to Solot, and his study was based en- 
tirely on sea-level pressure patterns. An 18- 
to 20-year, homogeneous series of mean 
monthly 700-millibar charts has now been 
prepared by the Extended Forecast Section, 
and it has been deemed desirable to develop 
a method based on this series, since it is the 
^ Published with the approval of the Director as 
Technical Paper No. 225 of the Pineapple Research 
Institute, Honolulu, Hawaii. Manuscript received No- 
vember 24, 1953. 
2 United States Weather Bureau, Washington, D.C. 
700-millibar level for which the basic prog- 
nosis is made (Namias, 1953). 
The present study is a continuation of a 
former study (Stidd, 1954) in which the cor- 
relation field technique was explored, tested, 
and compared with other methods. 
The main purpose of this study is to de- 
scribe the development of an objective meth- 
od for interpreting mean weather charts in 
terms of rainfall, although some attempt is 
made to supply physical explanations for the 
relationships revealed by the data. ^ 
METHOD 
The correlation field technique seems well 
suited to the problem of relating a parameter 
in two dimensions (in this case, the height 
of the 700-mb. surface) to a point parameter 
(rainfall), since any linear relationships must 
be revealed and use is made of all data. Fur- 
thermore, since no prior knowledge of the 
relationships is assumed, the method is ob- 
jective in its development as well as in its 
application. 
To prepare a correlation field pattern for a 
given rainfall parameter, it is necessary to 
determine the correlation coefficients between 
the given rainfall and the 700-millibar heights 
at each of a number of grid points. These 
coefficients are then plotted on a base map 
at the respective grid points, and the pattern 
is analysed. Such patterns usually are found 
to be smooth and regular with one or two 
264 
