FLOOD01375

• County: Larimer
• Title: Mesoscale and Radar Observations of the Fort Collins Flash Flood of 28 July 1997
• Date: 2/1/1999
• Prepared By: American Meteorological Society
• Floodplain - Doc Type: Flood Documentation Report

.

assessment report from the Department ,
of Commerce (NOAA 1997) suggests
that the KFfG 6-h STP maximum was
on the order of 6 in, (cf, KCYS and
CSU-cHILL estimates; Table 2),
In addition to Z-R estimates, the KCYS
CSU-cHILL data permitted a variety of
polarimetric rainfall estimates to be com-
puted using multiparameter variables in-
cluding KDP and ZDR (Doviak and Zrnic
1993; Ryzhkov and Zrnic 1995), For
brevity, only the polarimetric estimates
of STP computed using combinations of
KDP and KDR (Ryzhkov and Zrnic 1995)
and KDP' ZDR' and Z (e,g" a "blended
product") are shown in Table 2, Theoreti-
cally, the use of both KDP and ZDR in the
same rain-rate relationship (Table 2) should better
account for spatial and temporal variation in the drop
size distribution (cf, Ryzhkov and Zrnic 1995),
A subjective comparison of radar STP estimates to
rain gauge STP was conducted, The criteria for evalu-
ating the various radar techniques relative to the
gauges involved a comparison of the location and
value of the STP maximum and the overall pattern of
the rainfall. The R(KDP' ZDR) technique provided the
best match to the gauge STP using an equation previ-
ously cited in the literature, with no tuning, and a
minimum in processing (e,g" spurious data and ZuR
bias were removed), Two other polarimetric tech-
niques cited in the literature were also utilized [e,g"
R(KDP)' and R(Z, ZDR); cf, Doviak and Zrnic 1993] but
provided little improvement over the NEXRAD Z-R
relationship in an absolute sense, Detailed discussion
of radar-rainfall measurement using all of the poten-
tial polarimetric variables/techniques and their appli-
cation to this case is beyond the scope of this paper
but is the subject of several ongoing studies,
Examining the STP estimates in Table 2, the
NEXRAD Z-R applied to both KCYS (Fig, 19a) and
CSU-cHILL reflectivity data produced maximum
STPs of 5 and 6,5 in" respectively, only 50%-65% of
the gauge total (as was the KFfG estimate; NOAA
1997), Truncation of the reflectivity values to an up-
per limit of 53 dBZ, the maximum reflectivity used in
the NEXRAD Z-Ralgorithm, had little effect on the
calculated STP, The spatial distribution of the KCYS
(Fig, 19a) and CSU-cHILL STPs are broadly consis-
tent with the gauge analysis (Fig, 3b), However, the
CSU-cHILL STP maximum was located some 500-
1000 m southeast of the KCYS and gauge network

TABLE 2, Radar rainfall estimates of STP max (1725-2225* MDT),

Radar

STP Max (in.)"

Method

Z-R: Z ; 300R"
Z-R: Z; 139RL"
Z-R: Z; 250R "0

5,0
7,9
10,8

CSU-CHlLL

Z-R: Z; 300RL4
Z-R: Z; 139RL"
Z-R: Z; 250RLZO
R(~p' ZOR) = 52 KDpo.9t>zDR ...{I.447
R(K",. ZDR)/Z-R

6,5
10.2
14,9
8,6
8,0

*1725-2215 for the CSU-CHILL
**Gauge STP maximum 10 in.

Bulletin of the American Meteorological Sociely

STP maxima, The tropical Z-R relationships yielded
STP distributions and amounts that were similar to the
gauge analysis (Fig, 3b), However, the Rosenfeld et al,
(1993) Z-R relationship produced a marked overesti-
mate (factor of 1.5) of the area-integrated STP relative
to the gauge network using CSU-cHILL reflectivity
data,
The multiparameter R(KDP' ZDR) technique yielded
a maximum STP of 8,6 in" approximately 85% of the
gauge value, and a spatial distribution of rainfall con-
sistent with that of the gauges (Fig, 3b), Over the en-
tire coverage area, the R(KDP' ZDR) estimate was
approximately 10%-25% lower than the gauge totals,
Though the STP computed from the R(KDP' Zm,) rela-
tionship was reasonably accurate, several instanta-
neous rain-rate estimates at given grid points over the
5 -h duration of the event were contaminated by noisy
KDP and ZuR values in regions of moderate to high re-
flectivity (e,g" 30-50 dBZ),
To correct the R(KDP' ZDR) rainfall estimates for
sampling errors and noise at light to moderate rainfall
rates (:> 15 mm h-'), we created a blended rainfall prod-
uct that utilized the R(KDP' ZuR) estimate in moderate
to heavy rain when Z > 38 dBZ, a linearly weighted
R(KDP' ZDJ/Z-R estimate of rain rate in regions oflight
rain when 35 :> Z:> 38 dBZ,and pure Z-R estimate in
reflectivities < 35 dBZ andlor for data points where the
KDP or L;,R were below predetermined noise thresholds,
The noise thresholds for KDP and ZDR were determined
by examining collocated grid points of KDP' ZDR' and
Z. By visual inspection, the noise thresholds were con-
servatively determined to be -0.30 km-I for KDP and
-0,5 dB for L;,R' both thresholds occurring at a reflec-
tivity of 38 dBZ. These thresholds are similar to the

211