Laserfiche WebLink
<br /> <br />OJ089,o <br />. <br /> <br />. <br />. <br />. <br />. <br />.- <br />II <br /> <br />Analyses indicated that relationships between variables <br />on an annual basis were feasible. The two major classifica- <br />tions of variables considered are flow, and climate vari- <br />ables. Figure 3 summarizes the data available in each of <br />these classifications and their periods of record. Flow <br />data are from mainstem and side channel stream gages and <br />diversion records. Climatological data include temperature <br />and precipitation for both plains and mountain weather <br />stations. An additional precipitation variable, plains <br />precipitation, was synthesized from a Thiessen polygon and <br />existing precipitation stations. <br /> <br />Computational Method Used. <br /> <br />Annual regression equations were computed by the step- <br />wise multiple regression program (STEPR) contained in the <br />IBM Application Package resident on Colorado State Univer- <br />sity's CDC Cyber 172 computer. This flexible program can <br />select on the basis of statistics the important independent <br />(predictor) variables in a stepwise sequence or allow the <br />user to make specific choices of such variables. Both <br />options were used in this study. Initially, all seemingly <br />applicable variables were used to allow the computer to gen- <br />erate several equations (each with one more independent <br />variable than its predecessor). These results were then <br />analyzed to determine physical reasonableness and overlap <br />between variables (multicollinearity). Results were <br /> <br />-13- <br />