WMOD00419

• Title: Convective Rain Rates and their Evolution During Storms in a Semiarid Climate
• Date: 8/8/1984
• Weather Modification - Doc Type: Report

AUGUST 1984

DONEAUD, IONESCU-NISCOV AND MILLER,JR.

at greater ranges. Little effect was noted for strong
clusters. In calculating the rain volume, the 20 dBz
threshold was chosen in relation to the radar sensitivity
rather than to any meteorological reasons. After all,
the general goal was to find the A TI threshold that
gave the best rain volume estimates (independent of
R). If higher thresholds were used in RERV compu-
tations, the results would give the total rain volume
inside the variable threshold rather than the total
rain. If we kept raising the threshold for RER V
computations (25 --+ 30 --+ 35 dBz), we might get
better correlation with A TI but the result would be
of little use.
Frequency distributions of R for the 163 clusters
recorded in 1980 are displayed in Fig. 3. The number
of clusters decreases as the reflectivity threshold in-
creases, since the clusters with maximum Z values
equal to or a little above the threshold were subse-
quently eliminated (163 clusters for 20 dBz, 149 for
25 dBz and 117 for 30 dBz). The frequency distri-
bution for the 20 dBz reflectivity threshold is uni-
modal, with 79% of the observations in the 1.5-2.5
mm h-I average rain rate class interval. So, if a 20
dBz threshold is used to generate rain volumes through
ATI computations, an approximate 2 mm h-I average
rain rate would give acceptable results in almost 80%
of the cases. The frequency distribution for the 25
dBz reflectivity threshold may also be considered
unimodal, in a first approximation, with a stronger
right skewness appearing in the distribution. The

1605

lower relative frequency of the mode (48%) and the
slight right skewness show a greater scatter of the
data. The average rain rate is -4.5 mm h-1 and the
standard deviation - 4.4 mm h-I. Thus when the
whole spectrum of clusters is included in the analysis,
the 25 dBz A TI threshold seems to be less satisfactory
for rain volume computation through the A TI tech-
nique. A bimodal distribution and a much greater
dispersion of the data around the modes are clearly
visible in Fig. 3 for the 30 dBz reflectivity threshold.
The mode interval increases with the A TI reflectivity
threshold as a consequence of smaller echo coverages
for higher thresholds and thus the omission of a
significant rainfall contribution due to dBz values
< 30. Therefore, the 30 dBz A TI threshold is not
considered satisfactory for computing rain rates and
rain volumes, at least under the computation condi-
tions used in this work.
To determine the types of clusters generating dif-
ferences among the geometries of the distributions,
two-dimensional frequency plots were drawn for the
25 and 30 dBz A TI reflectivity thresholds (Fig. 4).
Here every row represents a rain rate distribution (in
number of cases) for a particular A TI interval. The
small ATIs are at the bottom of the figure. From Fig.
4A, it is clear that the right skewness and the large
scatter of the rain rates are generated for a 25 dBz
A TI threshold by the small clusters. Figure 4A shows
an average rain rate of -4.0 mm h-1 no matter what
A TI interval is considered, if the very small clusters

80

NDCMP SUMMER 1980 DATA

/\
/ .
,/ \
/ \
\

5 10
AVERAGE RAIN RATE (mmhr-')

FIG. 3. Relative frequency distributions of R for ATI calculated using different dBz thresholds.

60
-
':Ie
~
>
()
z
W
::l
0
w 40
a:
IL
w
>
i=
cc
....
w
a:
20

ATI REFLECTIVITY
THRESHOLD

- 20 dBz
-- -- 25 dBz
-.-.- 30 dBz

Ii
/ \
/ \
\
\
\
\
\
\
\
\
\
\
\
, /
I,
. ,
.- ,
"--

NUMBER
OF CLUSTERS

163
149
117

i

/
/

.-.-.-....... /
...... .
......./

--

15

>17