WMOD00399

• Title: WMO Training Workshop on Weather Modification for Meteorologists - Lecture Notes
• Date: 12/1/1979
• Weather Modification - Doc Type: Report

- 65 -

1) the precipitation detection and measurement problem maxi-
mizes in summer and minimizes in winter;

2) measurement erJrors are greatest for convective storms of
short duration and least for widespread steady precipita-
tion of long duration;

3) the sampling requirements (Le., number of gages/mi 2)
decrease as the sampling area increases;

4) the sampling rt~quirements are greatest for the light rains
and least for heavy rains.

Woodley et al (1975) found qualitatively the same results for Florida
precipitation. It is likely that similar qualitative results would be
found virtually anywhere in the world. However, there are certain to be
quantitative differences which must be considered in designing a precip-
itation network. Examples of the types of procedures one might employ in
determining raingaging requir~ments are presented in the next section.

Determining Sampling Requirements for Network Design

Minimizing errors in the measurement of precipitation is of
first priority in weather modification experiments. When employing rain-
gages, this means that gage de:nsity :Illust be sufficient to resolve the
precipitation entity that will be the subject of modification. Huff
(1970) approached this problem by calculating sampling error as a func-
tion of mean precipitation, storm duration, gage density, and the size of
the sampling error after grouping the data by season, synoptic storm type
and precipitation type. Approximate normalization of the data was
achieved through use of logarithmic transformations. Sampling error was
defined as the difference betw,een the best estimate of the true mean
rainfall obtained from the maximum density of raingages on each network
and the mean rainfall obtained with a lesser gage density. Huff worked
with two networks having sizes and maximum gage densities of 400 mi2
(1080 km2); 8 mi2./gage (22 km2./gage) and 550 mi2 (1485 km2.); 11 mi2./gage
(30 km2./gage) respectively and five years of record for each network.

The relevant results were obtained from a general equation of

the form:

logE = a + b 10gP + c 10gG + d 10gT

(1)

where E is the average samplinl~ error in inches, P the areal mean precip-
itation in inches, G gage density in mi2/gage, T storm duration in hours
and a, b, c and d are constants.

Warm Season vs. Cold Season Sampling Requirements

Huff obtained the logarithmic equation constants shown in Table
1 for the cold and warm season for a network of 550 mi2. (1485 km2.).
Standard errors for the various regression constants are included to in-
dicate the reliability of the c:omputed values.