9513 - Laserfiche WebLink

Home Browse Search

Analysis of Historical Discharge Records 7 the discharge rating to ensure the accuracy of the discharge values. It is common practice in the USGS not to apply shifts if the percentage difference between the measured discharge and the discharge rating is within 5 percent. These methods are described in detail in Rantz and others (1982a, 1982b). A large proportion of gaging stations currently (2003) operated, including the three stations along the study reach, are equipped with satellite telemetry devices, enabling access of real-time data using the Internet. These real-time data are con- sidered provisional (subject to revision) because (1) the recorded stage data might be in error due to equipment malfunc- tions or other factors; (2) the application of shifts may require modification as more information of gage operation is deter- mined during the year; (3) or other reasons. In most cases, real- time discharge data are reasonably accurate because the tech- nology, operation, and maintenance of real-time discharge sta- tions are well established. During the course of, and at the end of a water year, the recorded unit-stage data undergo extensive quality assurance to ensure that the computed unit-discharge data and the computed daily mean discharges, have the best possible accuracy. During the quality assurance, the shifts that were applied previously to the real-time data may be modified. When all of the quality assurance is completed after the end of the water year, the daily mean discharges are finalized and pub- lished. Unit data are not published, but for USGS stations, the unit data are permanently stored in the National Water Informa- tion System. Data computation for the Whitewater and Redlands-Canal stations follows the procedures just described, resulting in pub- lished daily mean discharges that are quality assured. The below-Redlands-dam station, however, is operated only for pur- poses of real-time water-rights administration and water- resource management, and no quality-assured daily mean dis- charges are available. The only available discharge data are the unit data. These data are subject to a larger degree of error than the quality-assured data for the Whitewater and Redlands-Canal stations. The CDWR makes discharge measurements at regular intervals at the below-Redlands-dam station to identify any dis- chargerating shifts; however, once applied to the real-time data, no retrospective shift adjustments (quality assurance) to the data are made. Analysis of Shifts Discharge measurements made at the Whitewater station during WYs 1995-2002 usually ranged from 1,000 to 5,000 ft3/s, with some measurements outside of that range (fig. 3). Zero shifts, with a few exceptions, were used most of the time, and percentage differences in discharge for the shifts generally ranged from -~ to +4 percent (fig. 3). The USGS hydrologic technicians responsible for the station consider the use of mostly zero shifts with some percentage difference between the discharge and the discharge rating as a better approach to records computation than variable shifts with zero percentage difference because the stage-discharge relation at the station is quite stable and has not changed substantially during the past several years (J.R. Sullivan, U.S. Geological Survey, oral commun., 2004). Discharge measurements made at the below-Redlands- dam station during WYs 1995-2002 usually ranged from 200 to 1,800 ft3/s, with a few measurements made at discharges less than 200 ft3/s (fig. 4). Shifts were variable throughout the period and always were negative; reasons for the large decrease in shifts during 1999-2000 are not known, but could be attrib- utable to channel changes resulting from construction of the fish ladder at the Redlands Canal diversion dam or changes in oper- ation of the sand trap at the Redlands Canal diversion dam. Per- centage differences in discharge, when applying the indicated shifts, were zero, with a few exceptions (fig. 4). Discharge measurements made at the Redlands-Canal station during WYs 1995-2002 usually ranged from 700 to 900 ft3/s, with a few measurements made at discharges less than 700 ft3/s (fig. 5). Shifts were variable throughout the period and ranged from -0.18 to +0.11 feet. Percentage differences, when applying the indicated shifts, frequently were zero, but about one-third of the measurements had a percentage difference other than zero (fig. 5). Although the percentage differences for the three stations (figs. 3-5) generally do not appear to be particularly one-sided, box plots were prepared to further evaluate the percentage dif- ferences (fig. 6). Based on the results shown in figures 3-6, it was concluded that application of percentage differences in dis- charge did not result in any discernible bias in the computation of the discharge records for the three stations along the study reach. Use of discharge rating shifts in computation of discharge records is a common and necessary practice and increases the accuracy of the record. In computation of a discharge record, use of percentage differences in discharge (difference between the measured discharge and the discharge rating) requires criti- cal evaluation. Use of percentage differences that have either negative or positive tendencies can introduce some bias into a computed record. Discharge rating shift and percentage differ- ence in discharge data were compiled for the three stations along the study reach and graphically analyzed in order to gain some understanding of the shift history. Characteristics of the box plots (fig. 6, and others to fol- low) are as follows (paraphrased from Helsel and Hirsch, 2002, p. 25): The median (50th percentile) is the midpoint of the data values. The end points of the box, the 25th and 75th percentiles, define the interquartile range (IQR). The whiskers extend to the furthest data point that is within one step beyond either end of the box (adjacent values); one step is equal to 1.5 times the IQR. Data points that are between one and two steps from the box in either direction are plotted with an "x" (outliers). Data points that are farther than two steps beyond the box are plotted with an "d' (far outliers).