WMOD00459

• Title: Snow Accumulation Algorithm for the WSR-88D Radar, Version 1
• Date: 6/1/1996
• Weather Modification - Doc Type: Report

Boucher and Wieler (1985) compared SD observations in cm h-1 against Ze measurements
made by a 3.2-cm radar for 6 Massachusetts snowstorms. Their Ze-SD relationship, when
converted to snow water equivalent by assuming the commonly-used 10:1 SD:S ratio and
using conventional units of mm h-1 results in:

Z = 227 S1.65
e

(6)

SD:S ratios vary considerably with snowfall type, degree of riming, and other factors.
Therefore, using an equation like (6) developed for SD may provide only a crude estimate of
S. For example, the a coefficient increases dramatically for low density snow. If the SD:S
ratio is 15:1, the a coefficient in equation (6) becomes 442; at 20:1, it becomes 711.

Smart and McGinley (1989) used hourly and 3-hourly observations of both SD and S from a
large volunteer network in the Boulder-Denver, Colorado, area. Radar observations were
made with a 10-cm radar with characteristics similar to the WSR-88D. Two case studies
were presented, and the measurements of SD expressed in cm h-1 were compared to a
reference relationship:

Z = 200 SD1.6
e

(7)

The a value in equation (7) would have the same value for S expressed in the usual mm h-1
if the SD:S ratio was 10:1. However, if the ratio was 20:1, the a value would be 606.

Plots of observed SD versus average Ze (in dBZ) showed that equation (7) rose about as
rapidly as the observations with increasing Ze in about the 25- to 35-dBZ range in both case
studies. However, the observations madle their steep exponential rise at about 5-dB lower
values than predicted by (7) in one case with "wet" snow averaging 12:1 SD:S ratios. In the
other case, with "dry" snow and SD:S ratios of about 25:1, the observations rose rapidly in
the 15- to 20-dBZ range, well lower than predicted by (7).

Plots of observed S versus Ze were compared to three similar reference curves including
equation (4). Smart and McGinley (1989) concluded that the reference relationships did not
provide enough reflectivity to satisfy the data. But no attempt was made to develop a better-.
fitting Ze-S or Ze-SD relationship with their data set.

Further comparisons with equation (7) were presented by Smart and Albers (1991) for two
upslope storms which affected a large volunteer network from Colorado Springs to Fort
Collins, Colorado, and far eastward onto the prairie. A 10A-cm radar was used, similar to
a WSR-88D. It was concluded that equation (7) in general underestimates snow depths by
about a factor of 3 or 4, and that the error became much worse for the case study with more
snow. Their results suggested that snow intensity in eastern Colorado is a strongly
dependent function of Ze between about 20 and 30 dBZ. Examination of figures 1 and 2
would thereby suggest a need for relatively low a and ~ values in equation (1). Further work
was recommended to better define a Ze-S relationship, taking into account ice crystal types,
degree of riming, and the SD:S ratio.

21