WMOD00283

• Sponsor Name: USBR, Technical Service Center, River Systems 7 Meteorology Group
• Project Name: Use of NEXRAD WSR-88D Radar Snow Accumulations Algorith in the GCIP LSA-NC
• Title: Use of NEXRAD WSR-88D Radar Snow Accumulations Algorith in the GCIP LSA-NC
• Prepared For: USBR Technical Service Center
• Prepared By: Arlin B. Super
• Date: 1/1/1998
• State: MN
• Weather Modification - Doc Type: Report

I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

are known to exist, the longer-term (e.g., storm total) VPR is expected to stabilize
and be at least somewhat similar from storm to storm. Koistinen (1991)
demonstrated that the VPR could be successfully used to range correct daily rainfall
accumulations out to at least 150 kIn from the radar.

To test these ideas with snowfall in the GCIP Large Scale Area - North Central
(LSA-NC) region, reflectivity (Ze) observations have been analyzed from the
Chanhassen, MN, WSR-88D radar (KMPX) made during the 1996-97 winter.
Archived Level II measurements (Crum et al. 1993) were obtained for the nine
largest dry snow storms. These data have been used to calculate the average
vertical profile of radar-estimated SWE for each storm.

Observations of the VPR in the lowest 3 kIn above the KMPX radar were used to
calculate SWE rates for each volume scan. The vertical profile of radar-estimated
SWE (hereafter referred to as VPS) has been calculated for the five lowest beam tilt
angles (about 0.50, 1.45, 2.40, 3.35, and 4.30 degrees) for all range bins from 33
through 37km range, and between 1 and 360 degrees of azimuth. This 5-km-wide
band of range bins was chosen to increase sample size. The mean range of 35 kIn
was chosen for a number of reasons. Perhaps most important, the calculations of
Andrieu and Creutin (1995) demonstrate that the apparent VPR observed by the
radar deviates from the true VPR depending upon range, the shape of the VPR, and
the characteristics of the radar beam. They show that relative errors in the
apparent VPR become serious beyond about 40 km range as smoothing of the true
VPR increases with range. For their hypothetical example containing a bright band
in the VPR, the bright band is practically undetectable beyond 70 km range.

The relationship Ze = 200 SWE2.0, where Ze has units of mm6 m-3 and SWE is the
instantaneous snowfall rate in units of mm h-\ was used to produce all calculations
and plots of radar-estimated SWE accumulations from Minnesota snow storms.
The SWE accumulation is the instantaneous snowfall rate integrated over many
volume scans. Radar reflectivity observations (in units of dBZ) were converted to
snowfall rate for each range bin sampled during each volume scan. This practice
eliminates the bias, contained in several reports and publications, caused by
averaging Ze or even dBZ values over time. The noted Ze-SWE relation may be
revised after further analysis of Minnesota data, but any changes are expected to be
limited and should not affect the results presented herein.

The above relationship should not be directly compared with the several snowfall
relations in the literature based on calculated radar reflectivity factor rather than
measured equivalent radar reflectivity factor. The former are based on snow
particle observations (Sekhon and Srivastava 1970) and result in coefficients that
are higher by a factor of about 4.5 (Smith 1984). Thus, the coefficient of 200 in the
above relation would be equivalent to about 900 ifbased on snow particle
observations rather than radar measurements.

3