WMOD00531

• Title: Raingage Network Requirements from a Simulated Convective Complex Weather Modification Experiment
• Date: 7/14/1980
• Weather Modification - Doc Type: Report

~i
f
.->;:;.
"If
~;.? <

.,} ~

'"r-;.
'5:
~
;.~:

1:

.
'.J~

,
'.

1182

JOURNAL OF APPLIED METEOROLOGY

VOLUME 19

1
1
I
\
! I

JS.ll

A

a---- _I'" 1Io1l11l'~1
t-~., IIo,HII' 511

A
,,~
. , A . . 100'4
a' .
i , . . 50'4
, B
I
J200 I
I B C . . 2~'4
I ...
I ,a ".A~C
, ,
J , . /
. ,'/C
80 G\,' .
60 L~'~~_c
40
'~......
'.
" ,C
'a
20 I'

300 :350 I

~

200

....... ., .....4 ,h.. 11_'. I_It.

FIG. 6. Numbers of CCs required to reach f3 = 0.10 for a
= 0.05 using data base stratified by size and three treatment
effects.

. S. Seasons required to detect a treatment effect

Each of the 1977 CC's were evaluated to estimate
which would likely have been operationaIly avail-
able had a seeding experiment been conducted. The
PPI presentations for each volume scan were
examined in chronological sequence for each day,
and the experimentally suitable CC's were manually
selected. To be a suitable experimental candidate,
the CC could neither merge nor leave the 150 km
radius of radar coverage within an hour of its first
echo. Further, candidates had to be separated by

TABLE 1. Operationally available convective complexes
at Miles City, MT. 1977.
--
---
Radar ~.~ope coverage \-iI Y4 ~ Full
Area ef coverage (km2) 8875 17750 35 500 71 000
Area rer gage for 250 gage
network (km2) 36 71 142 284
Convl.~tive complexes
available:
All 103 CC's 19 25 33 45
Smail half 13 18 24 34
Large half 13' 18 24 34

more than 2 h as it was assumed that only one CC
would be treated at a time, and that selection and
treatment would require 2 h. The resulting numbers
of CC's for the most active fractions of radar scope
coverage are shown in Table 1.
Interpolation from plots such as Figs. 5 and 6
yielded the numbers of CC's Irequired for the four-
gage densities listed in Tabll~ 1. Division by the
number of CC's operationaIly available during the
1977' season (assumed typical) resulted in an estimate
of seasons required to complete a seeding experi-
ment. For example, using the~ unstratified data set
(Fig. 5) and the probability levds a = 0.05, f3 = 0.10
with 8 = 100%, 10 seasons are required with 250
gages spaced at 36 km2 per gage, 10 also for 71, 8 for
142 and 6 for 284 km2 per gage. For these constraints,
the optimal density for a 250 gage network would be
284 km2 or more per gage; i.e., the gages should be
spaced over the entire area of radar coverage. A
summary of experimental seasons required for vari-
ous conditions is given in Table 2.
In those cases requiring ovc~r 350 samples for aU
spacings, only a minimum number of seasons could
be specified. This was more than seven years for the
entire 103 CC sample and more than ten for the
two halves.

6. An alternate approach

If the central limit theorem holds and the among-
storms plus sampling variances are assumed equal
for both seeded and non-seeded CC rainfall accumu-

TABLE 2. Numbers of experimental seasons required to detect ,Il-treatment effect and optimal areas over which to space 250 gages.
(Number seasonslarea expressed as fractions of 150 km radius radar coverage with E being the ent:ire area. I implies optimal
area indeterminant.)

1) = 25% o ,= 50% o = 100%
-~.~--
All Large Small AJ\ Large Small All Large Small
103 half half 103 half half 103 half half
",811 '" 1I11 '" II/I ",8/1 6/E 10/Y4 6/E 31E 6/Y4
",SII 81E 13/14 ",8/1 3/E 6/14 41E 21E 4/14
",811 3/E 41E ~811 21E 3/\4 liE lIE 21\4
"'811 I OlE 14M 11/'-: 41E 8/Y4 SM 21E 4/\4
11M 51';1 7M ~". 3/E 41'i4 31 'I.! liE 311
. _ Il~.
4/~ 21E 31E 1/f lIE 2/',:' liE liE liE
/

IX = 0.05

/3=0.10
/3 = 0.25
/3 = 0.50
/3 = 0.10
/3 = 0.25
fJ = 0.50

a = 0.10

. -: ~.,::-ptt:.;;:"t- .':/,;;,}:'-:-;;r~(I.. -.-L;-",'~;;';1. ;"'<:i:::~HZ.~(""" ~- f .-,,- "-"...<:

:.', .1,tc:, _,~-_.,:: _~"-, ~ .;, -' "._,,::, ~_ ~'...~ ,', ~.,,;.~. ", :", ",: ~<:W~'2:~r~r;'r" ~'~i"_ 7: ~ ._"."~\7-::~':~-~~."~~1!,~~~<<"!':,~ _ ,".;/:-';:- ,14-. ~, :'.;~ ,j,,," ~:;~. ;!,~~."r""'~;

.r,'..'~" ~