WMOD00312 ‎(2)‎

• Title: Cloud Seeding Feasibility for the Shasta-Trinity Watershed
• Date: 12/1/1993
• Weather Modification - Doc Type: Report

A month was considered missing if more than 91 d out of the month were missing. For
instance, COF (Coffee Creek) should have had 360 mo of data for the 30-yr record, but had
only 324. Using the percentages shown in table 4.2, one can derive the number of missing
months for the other stations. The final column shows the number of years having complete
data during the period October through May. One can see that very few years had complete
data except for SRA. Unfortunately these missing data do impact the results ofthis analysis
because estimates of hours of precipitation will be underestimated. Given the poor data set
for two of the stations, only the results from SRA and COF will be discussed.

The hourly data were provided in event format on optical disc. That is, only hours having
precipitation above 0.10 in. were included. The method by which these events was processed
is as follows. Events were grouped in what will be called STORMS. A STORM had to have
at least a 3-h duration and was considered ended if a 24-h break in precipitation occurred.
A list was then generated of each STORM by year..month-day and beginning hour, length of
STORM, amount of precipitation that fell, and if arlY of the hours were flagged. An hour was
flagged if a precipitation amount was accumulated because of several missing hours of data.
Thus, STORMS with no flags were called "good STORMS" and STORMS with flags were
included in a group called "all STORMS."

STORMS were also grouped by water year and amount. As discussed in section 4.1.3, the
daily and hourly precipitation amounts were derived from two separate, collocated gauges.
To check completeness of the hourly data, the total STORM precipitation for the months of
October through May could be compared to the precipitation totals for this same period from
the daily amounts. The two totals rarely matched. On average, the October to May
precipitation totals from hourly gauges were 88 pct of the totals compiled from the daily
observations. For the 17 supposedly good years from SRA, this percentage varied from 60
pct for the water year 1977 to 102 pct for the water year 1974. The explanation for this
discrepancy was determined by comparing the two data sets on a month-by-month basis. The
event data showed no precipitation on certain days but were not flagged as missing; the daily
data showed precipitation did fall. This discrepancy again reiterates that the STORM hours
computed from analysis of the event data will be an underestimate of the hours of
precipitation between October and May.

The number and size of STORMs were examined irrespective of water year. For the 41- year
record at SRA, a total of 866 STORMs occurred, lasting a total of 17,627 h. Of these, 782
were considered good (no flags) and lasted a total of 15,287 h. The distribution of STORM
totals by I-in. amounts for all STORMs observed at SRA for the 41-yr record is shown on
figure 4.11. The median STORM produces about 1 in. of precipitation. The modal class is
0.1 to 1 in. As STORM totals increase, the number of STORMs decreases asymptotically.
This same relationship was observed for COF (fig. 4.12).

The average duration of STORMs for SRA having precipitation totals between 0.1 and-23 in.
is shown on figure 4.13. The average duration of STORMs steadily increases until STORM
totals reach about 14 in. STORM totals continue to increase after this point, but the duration
of STORMs levels off at 96 h (4 d). The average duration of all "good" STORMS was 19 h.

The longest STORM observed at SRA was 161 h, beginning on November 25, 1951. It
produced 14.75 in. of precipitation. The wettest STORM observed at SRA during the period
examined was 23.4 in., which began December 18, 1964, and lasted 93 h. This STORM

27