WSP08546 - Laserfiche WebLink

Home Browse Search

tv <::> l-" tv Approach Techniques that are used for record extension include correlation with data from tne historical period of record, correlation with data from a similar site (base station), regional basin analysis, synthesis based on statistics for the historical record, and conceptual modeling. Correlation techniques were selected for this study because they are easy to apply and have proven effective in other studies (Hirsch, 1982; Alley and Burns, 1983). In addition, they are appropriate for the daily or monthly time scales of the streamflow and dissolved-solids data used in this study and for the type of record extension, which did not involve changes in drainage-basin characteristics. Four different correlation techniques were used to extend streamflow and dissolved-solids records in this study. All of these techniques were based on least-squares, linear regression. Streamflow at record-extension sites waS estimated by simple regression on data from a base-station site. Dissolved-solids concentration was estimated by weighted regression on streamflow, either at the record-extension site, if streamflow data were available, or at a base-station site. If a base-station site with water- quality data was available, dissolved-solids concentration was also estimated by simple regression on concentration at the base-station site. EXTENSION OF STREAMFLOW RECORDS Hirsch (1982) compared four techniques of extending streamflow records using correlation. Two of the techniques were based on least-squares regres- sion using data from a base-station site. The other two techniques were developed by Hirsch to maintain the variance of historical data at the ex- tension site. He called these techniques "Maintenance Of Variance Extension" types 1 and 2 (MOVE.1 and MOVE.2). In empirical trials~ Hirsch found MOVE. 1 and MOVE.2 effectively preserved certain time-series properties of the historical record, such as variance and extreme-order statistics. However, the goal of the present study was to produce the best estimate for each individual streamflow value. Regression is most suitable for this purpose, and therefore was the technique selected for extension of streamflow records in this study. Alley and Burns (1983) developed a method for selecting tne best base- station site if several possibilities are available. They achieved good results using regression, MOVE.1, and MOVE.2 analyses. However, in the present study selection of a base-station site was not considered a problem. There were few stations with long-term records during the period of interest at the record-extension sites, and there was always one base-station site that was clearly preferable. Daily values of streamflow at the five sites requ1r1ng streamflow-record extension and at their associated base-station sites were retrieved from the u.S. Geological Survey's National Water Data Storage and Retrieval System (WATSTORE). These data were used to make regression estimates of daily streamflow, in cubic feet per second, which were combined to yield monthly streamflow, in acre-feet per month. 5 -.u .