Methods


Analyses discussed in this report consist of an accounting procedure in which a gain or loss in streamflow, termed seepage, from one measurement site to the next site downstream is attributed to interaction with the groundwater system after compensating for known withdrawals and tributary inflows. A gain in streamflow indicates that groundwater is flowing into the stream through the streambed or by way of bankside seepage, whereas a loss indicates that stream water is flowing into the groundwater system through the streambed. Values of net anthropogenic withdrawals between two measurement sites are added to the measured streamflow at the downstream site, whereas intervening tributary inflow is subtracted from the measured streamflow at the downstream site. The result is the net gain or loss in streamflow from the upstream measurement site to the downstream measurement site. Values for tributary inflows were available during only some of the studies and, therefore, were sometimes estimated based on streamflow measurements made in tributaries during the same season in other years. Analyses are from the base-flow period, typically August–September, when streamflow is minimally affected by surface runoff from precipitation. 


The basic components of this seepage investigation were individual streamflow measurements and continuous records of streamflow. All data are available online at http://waterdata.usgs.gov/or/nwis/sw using the 8- or 15-digit USGS site number, as appropriate. 


Streamflow measurements made at numerous locations during a short (1–2 day) period enable the reconstruction of a synoptic “snapshot” of seepage. Streamflow measurement sites (the location of the actual measurement) and streamflow monitoring sites (the source of continuous streamflow records) are identified by site in table 1 and shown in figure 2. Measurement sites on the lower Clackamas River were associated with the river mile, measured from 2005 aerial photography (Oregon State University, 2008), and may differ in their river mile location from previously published reports and maps, primarily due to channel changes that resulted from a major flood in 1996. Streamflow was measured using mechanical or acoustical meters, following guidelines of Rantz and others (1982) and Oberg and others (2005), respectively. 


Seepage also can be interpreted on a monthly to seasonal time scale on the basis of streamflow records from USGS monitoring sites for the low-flow period each year. Continuous streamflow records are obtained by applying a “stage‑discharge” rating to automatically recorded stream levels (stage) to obtain a series of estimated streamflow (discharge) values. The stage-discharge rating for a monitoring site is developed from the relation of stream levels to manually measured streamflows associated with those stream levels (Rantz and others, 1982). Determining gains and losses from long-term streamflow record has advantages and disadvantages compared to the synoptic approach. The primary advantage is that the effect of hour-by-hour streamflow fluctuations on the data record is muted. This method is also cost-effective because it uses existing daily streamflow data collected for multiple purposes and does not require coordination of a team of hydrographers to measure streamflow at a given time. The primary disadvantage is that the long-term record includes unmeasured streamflow contributions from tributaries; as a result, main-stem streamflow cannot be mathematically isolated. Additionally, surface-runoff events sometimes occur during late summer. Depending on runoff characteristics of the basin upstream of Estacada relative to downstream, differences in streamflow might be incorrectly interpreted as gains or losses due to seepage. However, the effect of surface runoff is probably minimal due to infrequent summer precipitation and seasonally high evaporative losses.


The overall accuracy of a seepage investigation depends on the accuracy of the determination of streamflow difference from one measurement site to the next. The difference in streamflow, whether calculated from discrete streamflow measurements or continuous streamflow records, must be equal to or greater than the uncertainty of that difference to be considered significant. The estimation of streamflow uncertainty is based on ambient conditions and the methods used to measure the components of streamflow: stream velocity, channel depth, and channel width (Rantz and others, 1982). For this study, the range of uncertainty was 5–11 percent, depending on measurement method and stream channel characteristics. 


Continuous streamflow records have inherent uncertainty that is determined by the accuracy of the stage record and the reliability of the stage-discharge relation, which is determined in large part by the frequency of manual streamflow measurements to verify the relation over time and environmental factors, such as the stability of the stream channel. Continuous streamflow records are rated annually, and for Clackamas River monitoring sites used in this study ranged from “excellent” to “poor” on an annual basis, depending on the site, representing an uncertainty of 5 percent to greater than 15 percent of the true value. In this seepage study, however, low flows were of primary interest. Low flows generally can be estimated more accurately than flows during other times of the year because the streambed is likely more stable than during the high-flow season. Consequently, continuous streamflow records were assigned an uncertainty of 10 percent of the true value. 


Another source of uncertainty in a seepage investigation is the possibility of a rapid change in streamflow (pulse) in response to release changes from a reservoir. River Mill Dam (site 4, fig. 2, table 1) on the Clackamas River is the primary source of pulses in the reaches discussed in this report. Whereas a pulse is easily identified at the streamflow site immediately downstream of the dam (site 5) by an almost instantaneous change in streamflow, the translation of a pulse, in timing and magnitude, to the downstream sites is less certain. The travel time of flow pulses was estimated on the basis of continuous streamflow at Estacada (site 5) and Oregon City (site 32) (fig. 3). The time delay between a pulse in streamflow at the upstream site and its arrival at the downstream site was 5–8 hours for the range of streamflow discussed in this report. These results were used in a qualitative sense for the analysis of gains and losses of the Clackamas River and, particularly, to exclude comparisons of streamflow during times of increased (or decreased) streamflow due to changes in reservoir releases. 


Finally, reported withdrawals are also subject to uncertainty. For this study, withdrawals were relatively small compared to streamflow of the Clackamas River, and the reported values were incorporated into the seepage accounting using an estimated uncertainty of 5 percent. 


First posted November 28, 2011