Laserfiche WebLink
results. The summary file includes the independent stations that were used to calculate each missing <br />value and the pre- and post- regression population means and standard deviations. <br />Summary <br />Exhibits 8 and 9 have been provided to ease the comparisons between different diversion data extension <br />methods and data extension models referred to in this memo. Based upon the information within these <br />Exhibits and a review of the available data the following conclusions can be drawn: <br />• It is feasible to extend the historical record. Both climatological and streamflow data is available <br />back to 1950 for reasonably extending the current record, although those interviewed have expressed <br />a desire to see it extend back to the late 1920s. The record maybe extended this far, although more <br />uncertainty in the estimates would result because less gage and climatological data is available. <br />• The mixed station method appears to be the best method for data filling and extension for the <br />following reasons: (1) its ability to handle several gages with differing periods of record in a single <br />matrix; (2) its ability to use multiple independent series to fill one dependent series according to the <br />best correlations; (3) it requires the least amount of user discretion; and (4) the program is already <br />coded and public domain. <br />• As part of the data extension process, the accuracy of the extension can be estimated based upon <br />correlation coefficients and the accuracy of estimating natural flows for the independent stations. <br />Recommendations <br />Based upon this summary, the following recommendations are made: <br />• The natural flow data for CRDSS and StateMod should be extended back to 1929 to allow <br />comparisons to existing studies, and to provide the most complete data set possible for period of <br />record selection. <br />• For computation of natural flows, major diversions and reservoir operations data will be digitized <br />where available. Missing diversion data should be estimated by average diversions characterized for <br />average, wet, and dry year conditions using 1975-1991 data (based on streamflows). Municipal and <br />industrial demands will be adjusted based upon population estimates. Return flows will be estimated <br />based upon 1975-1991 average efficiencies. <br />• This data extension should be accomplished utilizing the mixed station method, requiring the <br />development of a user interface to fill the missing data. <br />• The user interface should be developed to compute basic statistics of synthesized natural flows to <br />compare with those of the 1975-1991 period. It should also estimate the accuracy of the synthesis by <br />comparing natural flows computed from recorded diversions to those computed from estimated <br />diversions for the 1975-1991 period. <br />Appendix E E-21 <br />