SEVEN YEARS OF WEATHER MODIFICATION IN CENTRAL AND SOUTHERN UTAII

A post-hoc statistical evaluation of an operational cloud seeding project, designed to enhance winter snowfall in the mountainous sections of central and southern Utah, was based on comparison between several control areas and multiple target areas. Linear regression equations were developed for each control-target area combination based on average January-March precipitation, 1956-1973; from 1974 through 1980 the target was seeded consistently during these months. Ratios of observed to calculated precipitation are presented for the target areas and several sub-target areas based on the predictions using the control area precipitation as predictor for the seeded years. Significance of the results was determined using a one-tailed WilcoxonMann-Whitney ranking test. On the basis of this evaluation it is concluded that seeding has been successful in increasing the January-March precipitation within the intended target areas over the seven year seeded period. The results vary depending on the control area, but all are positive. Indicated increases range from about eight percent to twenty-eight percent within the various target sub-sectors. Over the total Primary Target precipitation increases of between 13 and 20 percent are indicated. These results appear to be highly significant in most of the target areas. Some evidence of positive extra-area effects are noted "downwind" from the target, but with less statistical significance than the target areas.


INTRODUCTION
Utah is one of the driest of the 50 states; much of the precipitation that does fall accumulates as high elevation snowpack from October to April.For irrigated agriculture, important in the state, streamflow from these accumulated snowpacks is necessary, and any additional runoff that can be acquired is valuable.
An operational weather modification program in central and southern Utah began in the 1973-74 winter and has continued each winter through the present.
Sponsors of the program are the Utah Water Resources Development Corporation (formerly Southern Utah Water Resources Development Corporation) and the State of Utah, Division of Water Resources (DWR), active participant since 1975.
The program goal has been to increase winter snowpack at higher elevations of participating counties in central and southern Utah.North American Weather Consulants (NAWC), the weather modification contractor since the program's inception, has evaluated the effectiveness of the program for both five and seven years of operation.
The more recent 7-year evaluation of seeding of winter-spring seeding 1974-1980, is reported here.

BACKGROUND
The intended target areas of the program, as defined in an earlier design study (Thompson et al.., 1978), encompasses some 19,000 sq km (12,000 sq mi) 141 of mountains in central and southern Utah.Commissioners or water conservancy districts of counties potentially involved in the program vote annually whether to participate.
The DWR then shares the cost of the program.On the average, 12 counties .haveparticpated annually since the state involvement began in 1975 (Figure I).
Both silver iodide(Agl) ground generators and seeding aircraft were used in selected storm periods until 1979, after which only ground generators were used.
Manually operated units are supplemented by remotely controlled units at higher elevations.
For the current 1980-81 season, 75 manual and 4 remote generators were installed.
The program is strictly an.operational program designed to optimize any benefits, without any randomization.Portions of naturally occurring storms deemed seedable by established seedability criteria (Thompson et a|., 1978) are seeded.
Consequently, evaluation has relied upon comparisons of target and control precipitation.
The earlier evaluation of five years of seeding (Thompson, 1979) also was based upon comparisons of target control precipitation, but only from January to March.This was the only period consistently seeded during the five years.This five-year evaluation work was independently verified by the  ~e~ (ha~.ched), 1974-1980.Division of Water Resources (Div. of Water Resources, £981).Thompson (1979) found that the 1974-78 January-March precipitation in the Primary Target was greater than that predicted by the regression equations.Some differences between observed and calculated values were highly significant, particularlx in the southern half of the target.For the total target, the indicated increase due to seeding was approximately 18 percent, highly significant at the .018level by the Wilcoxon-Mann-Whitney rank test.

Both this evaluation
and that of the DWR were reviewed by a Technical Advisory Board organized by the DWR to provide guidance and expertise.Dr. Ruben Gabriel, Univ. of Rochester statistician, reviewed the five-year evaluation as part of a National Oceanic and Atmospheric Administration (NOAA) program to design a comprehensive evaluation of both the Utah and North Dakota state/local seeding programs. 142 The major recommendation of both the Technical Advisory Board and Dr. Gabriel was for additiomal analyses using other stations as control areas, since the evaluation used control stations selected post hoc.
The new analyses were suggested to confirm that comparable results could be achieved with different controls.
In ad4ition, Dr. Gabriel recommended elimination of four of the stations in the original ten station control due to their proximity to the target.These recommendations have been followed; the results of the evaluation, updated to cover seven years of seeding, 1974-1980 follows.

EVALUATION APPROACH
To cover sewen years of seeding the evaluation approach is the same as that of the five-year evaluation; e.g., statistical, using the correlation between a control area and a target area to determine the apparent effectiveness of seeding, Like the previous five-year evaluation, the basic data consist of January through March precipitation.The Eastern area is not in the intended target area, but was included to investigate potential extra area effects.The North Central area was selected to seek seeding effects in an area not expected to be directly affected by the seeding.In this manner, some assessment could be made of whether there was some bias in the seeded years favoring the likelihood of detecting a "seeding effect".
The isolated target areas in southeastern Uta.h were not evaluated in the five-year evaluation and have also been excluded from this evaluation due to lack of long-term precipitation stations.Precipitation gage locations are indicated in Fig. 2  Except for some storage gages, data were available back to about 1950.However, another seeding project had been conducted in southern Utah from 1951 to 1955 (Hales et al, 1955).To eliminate any possible effect from that seeding these years were excluded and the historical data base was formed from January-March precipitation 1956-1973.

Data Quality
The control area stations had consisent data, although occasional monthly totals were reported as estimated.Precipitation data were available in the Central target for 17 low elevation stations and 17 storage gages at higher elevations.
Less than five percent of the data were missing but, on those occasions when they were, estimates were made from surrounding locations by plotting the available data on charts and drawing isohyetals around the missing data areas.For the eight storage gages installed during 1956 and the one in 1957, data were estimated from surrounding data in a similar manner.The storage gages normally were read at or near the end of each month.When two or more months were combined, an interpolated value for the desired period was computed from the known percentage of precipitation which fell at the surrounding sites.
In the .South Central target area, the 12 low elevation precipitation gages all have consistent records.Data were available from five storage gages, although 1956 data were estimated at two.Less than three percent of the historical data were missing from the storage gages, and all was available in the seeded years.
The Dixie target area has five lower elevation precipitation stations, and two storage gages, at lower elevations than the ones in the Central and South Central target areas.These sites were not activated until 1959 and the data for the first three years have been estimated from surrounding locations.County), precipitation gages in the Salt Lake and Provo areas were not chosen.

Data Compilation
P~ecipiYation for each station within a group, either the controls or the various target groups, was summed for three months, January through March, for each individual year.These three monthly totals were then summed for all the stations within the group and averaged to obtain a "yearly" average.
Station histories of the precipitation gages indicate that many have been moved from a few to several hundred meters; elevation changes were generally less than 30 meters, but occasionally as much as 120 meters.
A few gages have been moved one to as much as five kilometers, but records at the nearest gages suggest no appreciable changes have occurred.
The storage gages' histories indicate a more stable pattern with little movement.Minor elevation changes (generally less than 60 meters) occurred at about half the sites.None of the moves was significant enough to change the station precipitation pattern and thereby affect the regression computations.
Where appropriate, target groups were stratified by elevation (valley precipitation gages and mountain storage gages).The year 1975 also indicates that little seeding effect, for reasons that are not as apparent.

5.
COWCLUSIONS Post hoc analyses using three separate controls to predict target precipitation from three sets of regression equations, while differing in indicated percentage increase, generally agree qualitatively.
All indicate that the January-March target area precipitation, 1974-1980(1976-80 for East Tooele), was greater than that predicted by the regression equations.Some ~ifFerences between observed and calculated (expected) values were highly significant.
In the five-year evaluation, Thompson (1979) noted that the southern portion of the Primary Target had higher ratios which achieved greater significance than those in the northern portion.
This trend has continued through the seven-year evaluation; the indicated increases in the south are about twice those in the north.The southern area indicates statistical significance about one order of magnitude greater than the northern portion, except for Dixie where several seeding suspensions during the last two years, might account for the lower significance level.

Evidence
of positive extra area effects continues to appear to the east of the target, but the results are not highly significant.North of the target, an indication of a minor seeding effect, is even less statistically significant than east of the target where carryover effects could be occurring for the last five years, from seeding in Tooele County,
Nevada and Utah, published by the National Climatic Center, NOAA, Storage-Gage Precipitation Data for western United States, also published by the National Climatic Center, and the Water Supply Outlook for Utah, published by the Soil Conservation Service, USDA.
Precipitation stations with reliable long-term records were sought in the Climatological Data for Arizona, 143

Table Lat .
Precipitation stations.

Table 2 .
Linear regression equations developed for target areas.

Table 3 -
Summary of correlations, ratios and significance for various sub-sectors of target