WMOD00440

• Title: HIPLEX-1: Statistical Evaluation
• Date: 4/4/1984
• Weather Modification - Doc Type: Report

520

JOURNAL OF CLIMATE AND APPLIED METEOROLOGY

VOLUME 23

1.0

_.--.--.
.-
_e""-
.-
I
/

0.1

w
:::>
..J
<{
>
ci.
UJ
~ ~.A---'1\
~ 0.01 t:r.... "\
>
>=
..J
:::>
::;;

0.001

RAW DATA

0---0 CLASS A-' CLOUDS

.--. CLASS B CLOUDS

an-n CLASS (A-' & B)
CLOUDS

0.0001 <Xl
'" U1 <Xl U
0 0:: UJ .0: a: a: a:
u u u > u. u. a.. a.. UJ UJ
U U 0:: ::;; f- I- >= >= .0: a:
RESPONSE VARIABLES

FIG. 3. Logarithmic scaled multivariate P-values based on MRPP
with commensurate raw data for primary response variables.

UJ
:::>
..J
.0:
>
ci.
W
f-
.0:
~ 0.01
>
>=
..J
:::>
::;;

0.0001

1.0

0.1

-.
.-_.--.-
.--
.....-
/

./
/
/
---
.

0.001

p,,-A
--
-~--,~'
,6--
,/fr---~"
--",'
",tt"

,J!'

,'~;'
"
I
,
,
d

RANKED DATA
0---0 CLASS A-' CLOUDS
.--. CLASS B CLOUDS
lr-un CLASS (A-l & B)
CLOUDS

'"
u
U

U1
a:
u
u

m
r

UJ
U.
I-

U1
U
U

.0:
a..
>=

a:
a..
>=

<Xl
U
0::

,0::
u.
I-

a:
UJ
<{

',;i,

RESPONSE VARIABLES
FIG. 4. As in Fig. 3, but with ranked data.

. SEEDED CASES
o NON-SEEDED CASES

8

\
\'

6

:.::
z
.0:
a::
'"
5 4
o

2

o

2

8

4 6
CIC2 RANK

FIG. 5. Illustration of the separation of points for the Class A-I
clouds in ranked-data 3-space.

(for most of the cases) with the addition of one or
another of the response variables measured at about
8 min after treatment. This observation is consistent
with the results of the univariate comparisons given
in Table 6.
These results are suggestive of a seeding effect dis-
cernible through the development of the ice crystal
plume up to about 5 min after treatment. They also
give an indication that the seeding effects in the chain
of events in the HIPLEX-I physical hypothesis could
not be followed past the development of the initial ice
crystal plume. It seems reasonable to inquire whether
the subsequent differences considered together are sig-
nificant. Table 8 presents multivariate comparisons of
the "post-5-min" primary response variables (PIC8
. through RERC). The high P-values indicate no obvious
differences between the nonseeded and seeded exper-
imental units for those response variables intended to
monitor the subsequent development of precipitation.
Univariate comparisons between nonseeded and
seeded experimental units are presented in Table 9 for
two secondary response variables (FCC5 and MAXZ).
The P-values indicate no obvious differences between

TABLE 8. PIC8 through RERC multivariate MRPP P.values for
"post-5-minute" response variables.

Cloud class Values P-value
A-I Raw data 0.257
A-I Ranks 0.286
B Raw data 0.880
B Ranks 0.768
A-I and B Raw data 0.528
A-I and B Ranks 0.584

u
a:
UJ
a: