Table 6. Frequency analysis of temperature dif-
ditions, and fewer than one-third of all differ-
ference along the river plane referenced to the
ences exceed this magnitude. About two-thirds of
differences exceeding 0.1C were observed when
temperature at Piermont bridge.
inversion was present over more than 10 cm of
Temperature difference (C)
snow cover. This analysis indicates that air tem-
Symbol
Distance
Under lapse
Under inversion
peratures measured within 100 m of the river
(Fig. 3)
(km)
median
84%
median
84%
may be indicative of the air temperature above
the river surface, but that differences may increase
h
0.1
<0.1
<0.1
<0.1
0.2
under inversion when the surface is uniformly
k
2.6
<0.1
0.3
0.3
0.8
covered with snow. Two measurements of air tem-
perature within 100 m of the river numerically
l
5.0
0.2
0.4
0.5
1.0
approximated six air temperature measurements
m
5.9
0.2
0.5
0.6
1.0
above the surface of the river. The differences are
within the resolution of the digital readout.
n
7.7
0.3
0.6
0.5
1.1
The difference between the surface tempera-
O
8.2
0.2
0.6
0.5
1.2
ture observed at the seven near river points and
the mean temperature measured along bridge P
u
29.4
0.6
1.2
1.0
1.6
has been used to construct a frequency analysis of
temperature difference as a function of distance
points along the river, and are listed in Table 6.
along the river plane. A frequency analysis of the
The lower case letter designators coordinate these
141 morning air temperatures observed on bridge
points with other places to be cited in Parts II and
P showed them to approximate a normal distri-
bution in the interval 2 > T > 26C.
III. The data for the point 100 m from the bridge
are repeated for reference in this table, but the
The temperature differences with respect to P
standard deviation is not calculated for this point
observed at the six points within 10 km of P were
in the following figures, as its 84th and 50th per-
stratified with respect to stability and pooled in
centiles lay within the digital resolution of the
two (lapse and inversion) data sets. When the
display.
logarithms of a variable are normally distributed
The median and geometric standard deviation
with respect to number of occurrences, the ratio
of the temperature differences at these seven near-
of the variable at the 84th and 50th percentile of
river points, with respect to the Piermont bridge
occurrence is equal to the geometric standard de-
(P) temperature, are plotted as a function of dis-
viation (GSD) of the variable. This method is well
tance south of P in Figure 7. The coordinator let-
described in Hinds (1982) and permits graphical
ter is noted on the distance scale. The 84th
extraction, or approximation, of the median and
percentile of temperature difference, and the cu-
geometric standard deviation from relatively
mulative frequency of lesser temperature at each
sparse data. About 40 observations under lapse,
point, are similarly plotted in Figure 8.
and 80 under inversion, are available for com-
From these analyses, air temperatures above
parison of temperatures along the river plane to
the relatively uniform surface of the frequently
the bridge temperature. Less than the total 158
recharged Connecticut River impoundment
data days are used in this analysis, as tempera-
shown in Figure 3 were found to differ by less
ture measurements along the northsouth extent
than 1C on most winter mornings. Critical analy-
of the experiment area were sometimes not com-
ses of Figures 7, 8 and Tables 4, 5, and 6 indicate
pleted before sunrise.
temperature difference systematically increases
Preliminary plots of the logarithm of the abso-
with distance from the reference bridge. Tem-
lute value of temperature difference versus num-
peratures within 10 km of P are quasi-uniformly
ber of occurrences showed that the logarithms of
distributed between colder and warmer with re-
the temperature differences were normally dis-
spect to that at P; those at 30 km are warmer than
tributed under lapse conditions. The logarithms
at P most of the time. These analyses indicate that
of more than 68% of the temperature differences
the plane intersecting the bank of the Connecticut
were normally distributed under inversion con-
River provides a relatively uniform reference
ditions.
plane, with respect to winter morning air tem-
The median and 84th percentile of tempera-
perature, along the segment of the Connecticut
ture difference have been extracted for lapse and
River Valley shown in Figure 3.
inversion conditions at the individual observing
10