Laserfiche WebLink
<br />c1, <br /> <br />Uncontrolled variance can result from many factors. Dennis (1980) stated that, "The <br />most serious difficulty with the historical regression method has to do with the stability in time of <br />the target-control relationship." For example, logging, insect infestations or fires may differently <br />influence target and control snowpack observation sites. Long-term regional climate shifts are <br />another possible source of uncontrolled variance. A recently published peer-reviewed article by <br />Super and Heimbach (2003) piOvides detailed discussions and examples of several potential <br />problems with the historical target-control method as it was applied in analysis of a four-year <br />operational project in southern Idaho. Their work illustrates how false interpretations of project <br />success can result. One of the major problems discovered was that frequent snowmelt during <br />March at some snow observing stations seriously compromised the original Idaho analyses. This <br />problem was largely eliminated by using March 1 st rather than April 1 st snow water equivalent <br />observations. <br /> <br />Examination of Colorado observations has revealed that snowmelt in March at some <br />stations will also affect historical target-control relationships, which use April 1 st observations. <br />The Super and Heimbach (2003) article addresses this problem in detail. As a consequence of the <br />many concerns with the historical regression method, any results should be viewed with caution <br />and considered as no more than suggestions. The method is incapable of providing "proof' in the <br />sense this term is used by scientists and statisticians. <br /> <br />A further source of uncontrolled variance is introduced when precipitation gauges are <br />used ineffectively to measure precipitation. Numerous investigations since the mid-1800s have <br />documented wind-caused under measurements of rain and much larger under measurements of <br />snowfall. The under catch is worse when gauges are not equipped with wind shields, and <br />increases with wind speed. But Groisman and Legates (1994) noted that the large majority of <br />gauges used in the U.S. are unshielded, and that even at first-order stations, only 20 to 40 percent <br />of gauges have been equipped with Alter wind shields the past several decades. Goodison (1978) <br />showed that significant under catch of snowfall results even with shielded gauges, when they are <br />exposed to light to moderate wind speeds. He demonstrated that the catch for Alter-shielded <br />gauges was about half of the true snowfall with a 10 mph wind at orifice height. Similar under <br />catches resulted with only 5 mph winds for unshielded gauges, and results were much worse with <br />stronger winds. These are very significant measurement errors, which are rarely correctable in <br />the absence of local wind measurements. Brown and Peak (1962) showed marked differences in <br />seasonal snowfall measurements at several Utah snow courses above 8,000 ft (all elevations <br />above mean sea level) compared with gauge observations at the same locations. The differences <br />were related to local exposure; that is, the degree of sheltering from the wind by forest and <br />surrounding terrain. Windier locations had lesser seasonal gauge accumulations than observed in <br />the snow course readings for April 1 st. <br /> <br />1.3 Overview of Winter Orographic Cloud Seeding Status <br /> <br />Research programs conducted on the Park Range, Grand Mesa and elsewhere in Colorado <br />have provided convincing evidence that supercooled liquid water (SL W) cloud frequency and SL W <br />flux over mountains do not appear to be limitations to seeding success. In other words, the basic <br />"raw material" needed for cloud seeding to augment snowfall is frequently present. If SL W clouds <br /> <br />6 <br /> <br />