WMOD00562

• Title: Snow Accumulation Algorithm for the WSR-80D Radar: Second Annual Report
• Date: 6/1/1997
• Weather Modification - Doc Type: Report

Table 5 summarizes the results of the new calculations. Only two Cleveland gages were used because
radar underestimation appeared to be a serious problem at the other three sites, from 87 to 146 km range
from the radar (see section 7.1). Four sites were used in the Denver area from 24 to 49 km from that
radar. Remaining Denver gages at farther ranges were also subject to radar underestimation because of
earth's curvature, beam broadening, and possible lack of beam filling by snow particles.

Table 5. - Summary of optimization technique applied to selected Cleveland and Denver data sets of gage-observed
and radar-estimated hourly snowfall accumulations showing resulting a and P values. Ranges/azimuths (kmldeg)
from the WSR-88D radar are noted in parentheses. Correlation coefficients, R, and average hourly snowfall
accumulations are given.

Ave Hourly
Location Pairs a P R SWE (inch)
Cle Gage I 300 417 2.2 0.61 0.019
(36/64 0)
Cle Gage 2 398 294 2.0 0.71 0.019
(61/690)
Gages I and 2 698 327 2.0 0.66 0.019
Den Gage 1 106 161 1.9 0.75 0.019
(25/2290)
Den Gage 2 100 137 2.0 0.71 0.025
(24/2630)
Den Gage 3 151 127 2.7 0.71 0.022
(49/2570)
Den Site 7 198' 199 2.3 0.72 0.014
(25/2650)
Gages 1,2 206 150 2.0 0.73 0.022
Gages 1,2,3 357 135 2.1 0.72 0.022
Sites 1,2,3,7 555 146 2.1 0.72 0.019
. 97 samples were trace observations set to 0.003 inch SWE

A comparison of the values will reveal that the R values of table 5 are less than those of the annual report
for all Cleveland gages but gage 5, located farthest from the radar. Furthermore, table 5 average gage
hourly accumulations were less for all five gages. These two findings are suspected to be related. The
lake effect influence diminishes with time as Lake Erie cools and partially freezes. This effect likely
resulted in less vigorous storms during the latter portion of the winter with lighter snowfall rates. The
expected weaker storms (on average) of late winter may have resulted in the smaller R values of table 5
because radar has increased difficulty observing shallow clouds with light snowfall rates.

The CTF for Cleveland's gage 1 varied only slightly between exponents of 2.0 and 2.2. Had the smaller
value been associated with the minimum CfF, the corresponding <<value would have been 380, still much
larger than the 294 found with gage 2. The relatively small R value betweeu gage 1 observations and
radar estimates overhead makes the optimized relationship suspect. Conversely, the Ze-S relation may
actually vary between gages 1 and 2 because of differences in orography and microphysics as speculated
in section 8. In the absence of other information, the recommended <<value for Cleveland is 330 based
on the two gages nearest the radar. This value is more than double that recommended for Denver, and
the differences appear to be real as will be discussed.

13