Jökull - 01.12.1969, Side 51

Fig- 3.
Correlation co-
efficients of an-
nual ice incid-
ence at Iceland
and Jan Mayen
temperature in
different
months.
FYLGNI MANAÐARHITA A JAN MAYEN
VIÐ HAFÍS Á ÍSLANDI ARID N
99% vissa fyrjr tengslui
<99% - -
ONDJ FMAMJ JAS
ÁRIÐ N-1
ONDJ FMAMJ J A S
Á R I Ð N
temperature and ice conditions off Iceland in
the following year. This might make it possible
to issue at the end of November an ice fore-
cast, valid until the end of September in the
next year. Tlie Jan Mayen temperature of this
period was therefore selected for further study.
With the aid of multiple correlation it was
found that in a forecasting equation for the
drift ice the temperature of the months Sept-
ember—November should be weighted about
50%
more than the temperature of the summer
months, June—August, in order to obtain the
°ptimum result.
Hereafter we call this weighted mean temp-
erature of June—November in Jan Mayen the
] M-temperature.
In Fig. 4 we show a scatter diagram depict-
Ing the correlation between the JM-tempera-
tnre and the ice incidence off Iceland in the
following ice year, counted in months. Unfor-
tunately (if one may say so!) there are very
few severe ice years in this period. One im-
portant dot may though be added, falling in
the middle between the two highest dots. This
corresponds to the ice year 1968/1969 which
was not included in the original study. It is
however evident, as could have been foreseen.
that there is not a linear regression between
the two factors, JM-temperature and ice in-
cidence. It was therefore desirable to find a
simple and reasonable mathematical model cle-
scribing this correlation. The following one
was selected:
I = 12/(10’X.T+b) + i) (Eq. 1)
where I is the anticipated ice incidence off Ice-
land in months, J is the JM-temperature, and
a and b are constants, determined by the ice
and temperature data. This seems to be a suit-
able model, since it makes the ice incidence
always more than 0 months and less tlian 12
months a year. The constant a describes how
rapidly the ice increases witli falling tempera-
ture and — b denotes the temperature that
corresponds to an annual ice incidence of six
months. The available data indicated that
a = 0.75 and b = 0.1. The relation correspond-
ing to these constants is drawn in Fig. 4.
JÖKULL 19. ÁR 47