WMOD00561

• Title: Snow Accumulation Algorithm for the WSR-80D Radar: Final Report
• Date: 7/1/1998
• Weather Modification - Doc Type: Report

Super (1998) applied the above approach to 9 Minnesota snow storms. His figure 1 shows that all
9 storms studied had approximately linear profiles of S in the lowest 2.7 km above the radar. With a
single exception, a linear least squares fit explained from 93 to 100 percent of the variance in the 5 data
pairs (each based on many samples) for each Minnesota storm. While slopes varied among storms the
median slope for all 9 was equivalent to a factor of 1.8 correction needed at 152 km range. Super applied
this "seasonal average" correction to all 9 storms at the single range of 152 km by way of example. He
demonstrated that corrected values ranged from 50 to 150 percent of presumed "true" values based on
extrapolation of individual storm profiles. The seasonal correction was within 20 percent of those
presumed "true" values for 6 of the 9 storms. This approach can certainly be considered a useful
correction scheme when one considers that failure to deal with the need for range correction for these
9 storms would result in underestimates, at 152 km, between 30-80 percent of "true" with most values
within 50-70 percent of "true." This finding is in general agreement with the results of Joss and
Waldvogel (1989) based on data from Switzerland.

A "first look" test was made of whether the near linear vertical profiles of radar estimated S found with
Minnesota storms is also found in other locations. Measurements of Ze for the durations of single storms
were chosen from the Albany, Cleveland, Denver and Minneapolis WSR-88Ds. The Grand Mesa
observations were not included because 35 km range is well beyond the Mesa top in most directions and
terrain blockage is severe for the two lowest beam tilts over the Mesa top. The only criterion for
choosing the particular storms was that they had significant snowfall. That is, they were not "hand-
picked" to attempt to prove a point.

Figure 3 shows the storm average vertical profiles for the 4 storms from different locations. Two lines
are shown for the Cleveland lake effect storm, the solid line based on an a of 260 (1995-96 winter) and
the dashed line on an a of 180 (November 1996 major storm). It is seen that this range of a does not
result in large differences in calculated S.

0.00
3000

0.02

0.04

0.06

SWE in. inches

2500

KFTG
(1 30)

.

500

\*
\
\
\
* \
KCLE
(180)

!...
o
-0
o 2000
a:::

Q)
>
..2 1500
o

VI
!...
21000
Q)
~

o
0.0 0.5 1.0 1.5 2.0
Radar-est. SWE (mm/h)

Figure 3. Sample vertical profiles of radar estimated SWE (snow water equivalent) for four radars using (3 =
2.0 and the a values in parentheses for the Ze-S relation. Linear regression lines are fit to the 5 data pairs
for each case. Both a = 180 (dashed line) and a = 260 (solid line) were used with the same KCLE data.

18