RESEARCH· ·1970·

years) in North America, Europe, and elsewhere in the
world. From statistical analyses of these records he
concluded, "There is no evidence that the climatic factors
as related to water resources have changed significantly
in the last 150 years." Furthermore, he found
that the effective annual precipitation (for the most
part annual runoff) is a random process.
I-Iidore (1966) computed the 50-year linear trend of
annunl runoff at 31 streamflow stations located
throughout the United States. From a study of his
work, a general nationwide trend in runoff was found.
Annual flood series for 45 streamflow stations in the
United Stntes with 43 to 75 years of record appear to
be trend free, according to Carrigan and Huzzen
( 1967). Similarly, the annual series of 1-dny and 7 -day
nnnual minimum flows for another set of streamflow
stntions in the eastern United States, with 45-47 years
of record, seem free of trends ( l\1atalas, 1963) .
PRIOR INVESTIGA'TIO'N.S OF PRECIPITATION TRENDS
l\1any investigations of trends in precipitation have
been reported. l\1ost studies limit consideration of trend
to the annual precipitation series observed at a few stations
in a limited region of the world. A few other
studies encompass analyses of trends in seasonal precipitation
or in storm rainfall.
Either annual rainfall, averaged over western Europe,
nppenrs to trend upwnrd between 1677 nnd 1940,
or nn nbrupt increase in precipitntion occurred about
1750; perhnps nn inhomogeneity in the records is indicated
(Brooks, 1954).
Smith (1963) studied sensonal trends in Great Britain.
The l\1arch rainfall at Stonyhurst was stationary
frmn 1851 to about 1910 and trended downward from
1910 to 1951, while that at l{ew was stationary for the
period 1851 to 1951. June rainfall in England was stationary
from 1730 to about 1830 and trended downward
from 1830 to 1950, while that at Wales was stationary
for the period from 1730 to 1950. The June­
,J uly rainfall for England and 'Vales definitely trended
downward from 1730 to 1950. The Octob~r rainfall at
l{ew trended downward for. the period from 1850 to
1950, and the November rainfall trended upward for
the same period.
No significant trends in either annual or seasonal
precipitation have been found in several areas of India
for records extending back to the 1800's ( Rao, 1963 ;
Rno and .Jngannathan, 1963).
Landsberg ( 1960) carefully selected stations in rural
areas of the United States, to insure homogeneity, for
a study of precipitation changes between the periods
1906-30 and 1931-55. Precipitation changes between
periods did not generally appear significant, although
CARRIGAN AND COBB
B215
a decrease from the first time period to the following
period may have occurred in areas of the Rocky Mountains,
Great Plains, and on the western side of the
Appalachians. At the same time an increase may have
occurred in the areas east of the Cascades, around the
Great Lakes, and in northern New England.
No statistically significant trends in long-term records
of precipitation appear to exist in Arizona and
western New l\1exico according to Sellers ( 1960) and
Hastings ( 1959).
Results of several investigations seem to indicate no
significant precipitation trends at the following locations
and periods of record: Rome, Italy (1751-1950);
Nicosia, Cyprus ( 1900-50) ; Beirut, Lebanon ( 1880-
1950) ; Alexandria ( 1870-1950), and Cairo ( 1890-
1950), Egypt; Khartoum, Sudan ( 1900-50) ; and the
Dead Sea (1815-1950} (Rosenan, 1963); Cape Town,
South Africa (1841-1960) (Hofmeyr and Schulze,
1963); Algiers, Algeria (1843-1962) (Dubie£, 1963);
and the U.S.S.R. (1841-1960) (Buchinsky, 1963).
REPORTED TRENDS rN TEMPERATURE
Many investigations, for example, those by Brooks
(1954}, Callendar (1961}, Dzerdzeevskii (1963), Willett
(1950), and Mitchell (1963} have been made of airtemperature
trends. l\1itchell's studies show that world
temperature trends were upward by about 1 °F, from
about 1880 to 1940, and downward since then. Changes
have generally been greatest in the northern hemisphere.
Mitchell indicates that trends have been worldwide
in scope and not confined to landmasses.
Such trends in temperature could possibly have some
effect on evapotranspiration rates. However, the temperature
trends that have been observed are so small
that the effects of slightly changed evapotranspiration
rates probably do not have a significant effect on runoff
patterns.
TREND ANALYSIS OF SELECTED PRECIPITATION
RE'CORDS
Regression ruralysis, use of moving averages, 'aJld use
of correlograms are techniques fur discerning the presence
of trends. If linearity of trend is assumed, the
t-test is used herein to determine if the regression
coefficient, representing the trend, is statistically significant.
(See Dixon and M·assey, 1957, p. 194.)
The computed trend is only an estimate of the true
trend. The true trend has a specified probability of
lying somewhere between the confidence limits 0+ and
0- as defined by the t-statistic. If the confidence band
lying between 0 + and 0- is small, then the magnitude
of the trend can be estimated with assurance ;