WMOD00459

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

hour of interest, and a simple average accumulation was calculated for the hour. Occasional
hours had scans from both clear air and precipitation mode scan strategies, representing
about 10- and 6-minute samples, respectively. But weighing each volume scan by the time
interval it represents would have little influence on the results because hours rarely had
different scan strategies.

Because each precipitation gage chart was read at the end of each hour, the radar snowfall
estimates were for the same hours. Future work will investigate whether any significant
improvement results from shifting the hourly radar-estimated snowfalls forward to allow time
for the snowflakes to fall from the 0.50 beam to the surface. The present results are based
on zero time lag between the two observational sets.

Future work will partition storms by types (upslope, lake effect, general synoptic, etc.) and
other means. However, the results herein are based on all available data extracted so far
with no partitioning. A few minor storms were ignored at both Cleveland and Denver if they
were of limited duration (few hours) and if hourly totals did not exceed 0.02 inch. Many
other storms provided an abundance of hours with accumulations of 0.02 inch h-l or less in
addition to higher accumulations.

Because gage charts were reduced at the "top" of each hour, hourly snowfall accumulations
are for 1200 to 1300, 1300 to 1400, etc. UTC on a 24-hour per day basis. Individual radar
volume scans were extracted for all the same hours for which gage charts were reduced; that
is, from the beginning to the end of any detectable snowfall by any gage during a storm
episode. This approach resulted in many hours with zero detectable snowfall by some or all
of the gages. However, unless otherwise stated, only hours with detectable snowfall (at least
0.005 inch) were used in the analyses to follow. This approach ignores many hours with light
radar-estimated accumulations. Such hours may have had virga, snowfall that missed the
gage, or snowfall that was too light to be detected by the gage.

In general, WSR-88D antennas never stop except for problems, maintenance, etc. When a
volume scan is completed by making the highest tilt scan, the antenna is moved to the lowest
elevation tilt and the next volume scan starts at 0.50 tilt. Volume scans do not take exactly
5, 6, or 10 minutes as suggested by Federal Meteorological Handbook No. 11 (1991), but
slightly shorter times, like, 5.75 or 9.67 minutes, which might vary from scan to scan. Scans
were separated by several seconds as the antenna moved from highest to lowest tilt. As a
result, most hours had from 9 to 11 volume scans with the precipitation mode scan strategy
usually used during snowfall. Ai:, few as 4 volume scans were accepted as adequate for the
occasional hours where continuous scanning was interrupted for whatever reason. Four
volume scans require about 23 minutes in precipitation mode scanning and about 39 minutes
in clear weather mode scanning, which is sometimes used at. the beginning and ending
portions of storm periods. But the large majority of hours had continuous radar coverage.

7. Z-S and Ze-S Relationships for Snowfall

7.1 Introduction

The fact that sensitive radars can detect and track precipitation echoes is of major importance
to weather forecasting and other weather-related activities such as aircraft operations. But
use of well-calibrated radars is also important to provide quantitative precipitation
accumulation estimates over large areas. To do so clearly requires that a relationship be

18