Jökull - 01.01.2005, Side 142
L. A. Rasmussen
Figure 2. Variation with wind direction of mean
October-April conditions over 1958–2003 at NCEP-
NCAR Reanalysis gridpoint at 65◦N, 17.5◦W: (a) rel-
ative humidity RH at 850 hPa, (b) temperature T at
800 m, (c) wind speed at 850 hPa, snow flux f
w
at 800 m, (d) frequency distribution of direction of
850 hPa wind. – icelandic
Here rms is the root mean square of the differences
between the model estimates b∗ and the measured val-
ues b, and σ is the standard deviation of b. Results are
shown in Table 3, along with those from two other
models.
When only two variables are involved (Tables
1,2), just r can be obtained, which shows by its sign
whether the variables are positively or negatively cor-
related. When more than two variables are involved,
such as in Equation (5), only r2 can be obtained. Cal-
culation of the statistical significance of r is described
in Chapter 7 of Bevington (1969).
Variation of critical direction φ′ is shown in Fig-
ure 4. For the glaciers in the northwest it is from the
southwest and for those in the northeast it is from the
southeast, rather than from over the ice cap. A similar
variation of the critical direction of the upper-air wind
occurs for stations surrounding the Olympic Moun-
tains (47.5◦N, 123.5◦W), as shown by Figure 1 of
Rasmussen et al. (2001).
Figure 3. Model (line) and observed (circles) seasonal
mass balance components. (a) winter balance bw, (b)
summer balance bs. The number to the left of each
curve is the value in m/a w.e. of the first point. The
number to the right of each curve identifies the glacier
(Table 1). – icelandic
Model results are only weakly sensitive to values
of model parameters z1 and z2 as well as to the pe-
142 JÖKULL No. 55