Laserfiche WebLink
with the climate. Review of data from agricultural uses in the basin does not seem to support this <br />assumption. This method also makes the assumption that irrigation cropping and other diversions are <br />constant over time. Although adjustments can be made if information is available, it lends another level <br />of cost and complexity to the data extension effort. <br />Finally, digitizing the data would be the best method, but the level of effort and the viability of early <br />diversion data makes this the most complex and costly method to implement. <br />The second method, characterization of diversions stratified by average, wet, and dry years and then <br />average monthly diversions for those years, is recommended as the best way to characterize diversion <br />demands. Actual data will be used for reservoir operations and trans-basin diversions where available. <br />Where data is not available, average, wet, and dry year characterizations of these operations will be <br />used. Generally, for larger reservoirs end of month content information is available although it is not <br />digitized. <br />Estimating Depletions <br />It is necessary to estimate depletions in order to deduce return flow data which will be lagged from the <br />actual diversion. Currently, StateMod uses an average efficiency for estimating the depletion of <br />irrigation water. Another method would be to estimate consumptive uses for the extended data set based <br />upon actual climatic data where climatic data exists. Where they do not exist, average, wet, and dry year <br />data could be characterized and averages for each type of year used. Municipal and industrial <br />consumptive use will be estimated based on historical data and relationships to industry and population <br />growth. <br />Potential Periods of Record <br />A review of the streamflow and climatological data available for the various basins indicates extension <br />of the record back to 1950 is feasible because in most basins a significant number of gages have records <br />back to approximately 1950. However, the extension of data back to the 1930s, which would include <br />another significant drought period, was expressed as desirable. The number of active gages existing <br />prior to 1950 is significantly less and the availability of climatological data is also reduced. Therefore, <br />the extension of data on all basins back to the late 1920s or 1930s is technically feasible, but its use may <br />be limited in terms of accuracy. An exception is the Colorado basin where more data may provide for <br />reasonably accurate data extension. <br />Regression Model Review <br />Several statistical packages and models were evaluated for data filling and extension capabilities applied <br />to the CRDSS database. Three suitable approaches were found: multiple regression, multivariate, and <br />mixed station. All are based on basic regression techniques, but differ in the approach taken. Following <br />is a discussion of the three methods, associated advantages and disadvantages, and a final <br />recommendation for a data filling/extension model to be used in the CRDSS. <br />The first method, multiple regression, is the most common data filling technique used. One regression <br />equation containing correlation terms for several independent flow series is used to fill or extend data in <br />the dependent series. The best set of series to be included in the regression equation (those most highly <br />E E-18 <br />