WMOD00459

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

established between what a radar measures, Ze' and R, the rainfall accumulation, or S, the
snowfall accumulation.

Many studies have been published concerning Z-R relationships for rain, usually of the form
of equation (1) with the a and ~ ,coefficients empirically determined, using units of mm6 per
m3 for Z and mm h-l for R. For a variety of reasons involving microphysics, vertical air
motions and other factors, no unique relationship exists for either rain or snow. Progress has
been made in optimizing values of a and ~ for different types of rain and different
geographical regions. However, the CUITlmt practice with the network of WSR-88s is to use
a = 300 and ~ = 1.4 as the default values in equation (1) for rainfall estimation nationwide.

Less work has been done with Z-S relationships for snowfall. Reasons for more emphasis on
rain include greater importance of rainfall to society in general, and the increased difficulty
of detecting snowfall because of its generally lighter rates, typically producing melted water
equivalent values in the range less than 0.10 inch h-l. Such accumulations are well below
common rain accumulations, especially from convective storms. Therefore, snowfall requires
sensitive radars for detection at even modl~rate ranges. Moreover, because the signal-to-noise
ratio is often relatively low for snowfall intensities, and because snowfall often develops in
shallow clouds, ground clutter returns can more easily hinder attempts to quantify snowfall.

Discussing the merits of various ground dutter suppression schemes is beyond the scope of
this report. Section 3.3 discusses the diffil~ulties of not having an automatic electronic record
of how, when, and where ground clutter suppression was applied with WSR-88D radars. It
is certainly recommended that steps be taken to record suppression methods in use on future
Level II tapes. The general approach taken in development of this Algorithm has been to
attempt to measure S in areas around Cleveland and Denver which presumably have little
or no ground clutter, even when that approach meant locating snow measurements farther
from the radar than desired for development of Ze-S relationships. The alternative was to
measure S within areas shown to have potential ground returns by the clutter bypass maps.
This practice would result in Ze observations that may have been suppressed to unknown
degrees, depending on the radial wind speed, the degree of suppression applied (notch width
map setting), and other factors. That approach could only add unknown variance to
development of a Ze-S relationship because the "true" Ze could never be determined.

Two basic, but fundamentally different, approaches can be used to relate radar measurements
to precipitation in general, and snowfall in particular. With the'most common approach, the
reflectivity factor is calculated from surface observations of snow particles without
involvement of radar measurements. The definition of the reflectivity factor, Z, is the
summation of the sixth powers of melted drop diameters divided by the contributing volume
Vc; that is:

Z:=LD6/Vc

(2)

This approach has a number of problems, including uncertainties in the density and fall
speeds of individual snowflakes needed to estimate the volume each particle represents.
These uncertainties are especially acute when often-present aggregation and/or riming are
important to the snowfall progress. Snowflake densities and fall velocities are usually
. estimated from empirical functions with particle size, introducing the uncertainty of how
representative these functions are for the particular snowfall being sampled. . Another
problem is obtaining sufficient particle observations to be representative of the large volumes

19