Laserfiche WebLink
apparent. This period begins around 1950 and continues into the early 1980s, for a total record <br />length of approximately 30 years. <br />Selection of the periods of record for those studies included in Exhibit 2 were based primarily on <br />the availability of historical data and the unique needs of the particular study. These needs included <br />the selection of a POR that would be consistent with that used in a previous study, inclusion of a <br />particular data set or type (i.e., drought periods), and public perceptions of the study period chosen. <br />The desirability of using later records because of probable higher accuracy, and the results of <br />statistical review were also influential in some cases. Little discussion was found on methods of <br />selecting the length of record for synthetic data generation. This is assumed to be dependent upon <br />the particular needs of the study being performed. <br />Once a study or synthetic period of record was selected, data was usually filled, extended, and/or <br />generated by one or more of several different techniques. These included cross-correlation, auto- <br />correlation, single and multiple regression, aggregation/disaggregation, physical and simulation <br />models, mass balance calculations, the use of surrogate data, and derived distributions. Discussion <br />as to why one method may have been chosen over another was limited. It is assumed that <br />preliminary statistical analysis of the database or the availability of additional information led to the <br />use of a particular technique. <br />Virgin or natural flows (those before the impact of man) were obtained before any stochastic or <br />modeling efforts were performed in most cases. These flows were calculated by backing out the <br />effects of diversions, imports, reservoir operation, consumptive uses, and return flows from <br />historical gage data. In the cases where natural flows were not calculated (Exhibit 2), either <br />stochastic analysis was not performed, or the model being used assumed that gage data from <br />streams that were higher up in the basin represented virgin flows, while those that were lower down <br />represented net reach gains or losses computed by mass balance techniques from the gage data. <br />Interviews with CRDSS Users <br />Interviews with various users of the CRDSS were held in order to assess the needs for data <br />extension and stochastic data generation within the CRDSS. Interviews were held with staff from <br />the Colorado Water Conservation Board (CWCB), State Engineer's Office, Colorado River Water <br />Conservation District (CRWCD), Denver Water Department (DWD), U.S. Bureau of Reclamation <br />(USBR), Northern Colorado Water Conservancy District (NCWCD), U.S. Fish and Wildlife <br />Service, and Harris Water Engineering, Inc. (southwest Colorado). Those interviewed included: <br />CWCB: <br />Randy Seaholm <br />Gene Jensock <br />Ray Alvarado <br />State Engineer's Office: <br />Ray Bennett <br />Bureau of Reclamation: <br />Randy Peterson <br />NCWCD: <br />Andy Pineda <br />U.S. Fish & Wildlife Service <br />George Smith <br />Appendix E E-3 <br />