WMOD00419

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

1604

MONTHLY WEATHER REVIEW

VOLUME 112

The rain volume was computed using an optimized
Z-R relationship, Z = 155R1.88, developed for this
region (Smith et al., 1975). The echo areas were
calculated for 20, 25 and 30 dBz thresholds corre-
sponding to 0.79, 1.04, and 2.70 mm h-1 rain rates,
respectively.

3. The average rain rate R

The value of R for a cluster is given by the ratio
of the total RER V to the A TI,

R = RERV/ATI.

The A TI is the echo coverage for a given radar
reflectivity factor threshold (i.e., 25 dBz) integrated
over the storm duration. If the RER V is given in
km2 mm and the A TI in km2 h, their ratio is in
mmh-I.
The radar scanning frequency was nominally at 10
min intervals; occasionally the intervals reached 15
to 20 min during hectic cloud seeding operations.
Beginning and end times of storms were determined
using a half scan interval (nominal value 5 min).
Scatter plots comparing the cluster rain volumes
and the A TI values expressed on logarithmic scales
show strong correlation between those variables (Do-
neaud et al., 1981, 1984; Ionescu-Niscov, 1982).
Linear relationships on such log-log scatter plots
imply a power-law relationship

RERV = K(ATI)b,

where K and b are regression parameters. From (1)
and (2) we get

R = K(ATI)b-l. (3)

As a result of Eq. (3), if b < 1 the average rainfall
rate would tend to decrease as the A TI increases. In
this case, larger, longer-lasting storms would give
lower average rainfall rates. This does not agree with
the commonly held notion concerning the behavior
of convective precipitation. If b =='1, the average
rainfall rate is independent of the A TI and R is
numerically equal to K. If b > 1, the average rain
rate increases with the A TI. As a result of (3), R
depends on the A TI (except for b = 1).

4. Climatology of the data

The summer of 1980 was climatologically dry in
the NDCMP region. The dryness and radar problems
resulted in only 163 usable echo clusters in 1980
from both radar sites. In the wetter summer of 1981,
587 clusters were identified usirig only Bowman radar
data. Cumulative frequency distributions of the cluster
durations were computed separately for the 1980 and
1981 data sets (Fig. 2). For both data sets, the
distributions were basically unimodal with the mode
being about 0.5-1.0 h and the median a little higher

8.0
1.0
6.0
5.0

~ 4.0
=-
z 3.0
o
;::
0(
~ 2.0
o
a:
w
....
Ul
::>
d 1.0

. JUL Y.AUGUST 1981 DATA

o JUNE.AUGUST 1980 DATA

(1)

0.5
1 2 5 10

80 90 95

99 99.8 99.9

50

CUMULATIVE FREQUENCY (%)

FiG. 2. Cumulative frequencies of the cluster duration for the
1980 (163 clusters) and 1981 (587 clusters) summer seasons
(NDCMP). Bowman and Parshall data were considered in 1980
but only Bowman data in 1981.

(2)

than 1 h. For the 1980 data, the longest cluster
duration was a little over 4 hours. The durations for
the wet 1981 summer were longer, with echoes lasting
up to more than 7 hours. This is emphasized by the
difference between the two lines in Fig. 2.
The frequency distribution of the times when the
clusters were first detected by radar showed an overall
first appearance of clusters any time during the 24
hours.

5. R as a function of A TI reflectivity threshold

As mentioned in Section 3, R calculated over the
lifetime of a storm depends on the A TI which, in
turn, is a function of the choice of reflectivity threshold
for estimating area coverages. The selection of the
most appropriate reflectivity threshold (MART) for
average rain rate computation depends partly on
subjective elements, but some objective factors are
involved. The MART is defined here as the reflectiVity
threshold of those three that were investigated which
gives the most accurate rain volume (RER V) com-
putation through the A TI technique. "Most accurate
RER V" means here the RER V which is the most
independent with respect to R.
Thresholds of 20, 25 and 30 dBz were used for
ATI computations. The total RERV was always
calculated using a 20 dBz threshold, no matter what
threshold was used in the A TI computations. This
approach needs discussion. The inconsistency in using
thresholds was related here to the study's goal. The
intention was to seek a method for estimating the
total cluster rain volume. To calculate this, all of the
reflectivity data, with no threshold, had to be used,
except that the radar sensitivity varied with range.
For weak clusters close to the radar, higher rain
amounts would result compared to the same clusters