WSP12615 (2) - Laserfiche WebLink

Home Browse Search

002537 ylay. with limited releases in June and July to preserve late-season release t1exibility to accommodate forecast increases. The monthly Glen Canyon release capacity was assumed to be about 1.9 MAF. The method of scheduling winter and spring monthly releases needs to be clearly explained, In attempting to model our operation to avoid spills, several degrees of aggressiveness in scheduling these releases were analyzed. In other words, if the forecast changed as the operation progressed through the winter and spring, this analysis attempted to answer the question ;'How should our operation react to these forecast changes?". In the past, an aggressive reaction to forecast changes was the key technique we used to reduce spill frequency, For example, if the forecast increased by 2 MAF at the beginning of April, then releases in April would immediately be increased, perhaps to maximwn capacity, in order to preserve as much t1exibility for future months as practicable. By reviewing the results of these computer runs, we then determined the frequency of unavoidable spills as well as the circwnstances under which a BHBF would be scheduled, all these under a range ofBHBF-triggering and "forecast-aggressiveness" asswnptions. Alternative BHBF Decision Criteria The subgroup considered several methods that could serve as threshold triggering mechanisms for implementing BHBF's, These include (I) various levels of risk associated with forecast error curves previously discussed with the TWG, (2) monthly volwnes, and (3) runoff forecast percentages. Each of these methods attempts to answer the question, "Under what conditions should a BHBF be triggered?". As these alternatives were analyzed, an attempt was made to eliminate operational errors in reacting to runoff forecasts, errors of omission as well as commission, An omission error would be to not release a March BHBF when a spill actually would occur later in the year, accompanied by high powerplant releases. As an example, Graph I shows the evolving spring runoff forecast during water year 1985, The drop in the forecast on March I would have dissuaded the release of a BHBF, when in actuality the dam would have spilled later in the spring when the inflow was much higher than expected, A commission error is the release of a BHBF when a subsequent spill would not have actually occurred, Graph 2 shows this type of situation in water year 1972, where the runoff forecast was originally high early in the winter, but dropped dramatically in the spring as the result of dry climatic conditions, A range of forecast risks from 2 to 50 percent were investigated. A vety liberal risk level of 2 percent recognizes all actual spill years but also allows BHBF's in many years in which there was no justification for such. The years 1970 and 1971 are good examples of such and had only slightly above normal runoff with very little risk of spill. Requiring high levels of risk as a threshold more accurately handles the non-spill years, but fails to identifY many of the actual spill years. Because of the complexity in evaluating spill risk using forecast error curves, it was suggested,that torecasti'should be evaluated from a "percent of normal" basis. The subgroup found that such an approach was actually an alternate expression for spill risk, but more straightforward and simpler to employ. Various percent of normal thresholds were evaluated, 5