Surface-Water Hydrology

The following section is a discussion of the surface-water hydrology of the Johnson Creek basin as it relates to trends in annual flow, low flow, and high flow. Analyses of trends include comparisons of streamflow data on Johnson Creek and data from a streamflow measurement site in a nearby, undeveloped basin to ascertain the effects of development on streamflow. The Little Sandy River (map number 45) is not affected by regulation, and the long period of continuous streamflow record allows long-term comparison of the period of record of Johnson Creek at Sycamore (map number 30), where streamflow data collection began in 1940. The two basins additionally are similar in size, shape, and orientation and are only about 20 mi apart. Land cover in the Little Sandy River basin primarily is forest, and periodic logging over the past century or more. However, hydrologic change because of land use and other human intervention generally has been less in the Little Sandy River basin than in the Johnson Creek basin.

Differences in both low and high flows are discussed in the sections below. The key differences are likely because of differences in elevation and precipitation. The elevation of the Little Sandy River basin is from about 700 to 4,300 ft, compared to 50 to 1,100 ft in the Johnson Creek basin. These differences result in lower temperature, more snow, and a more attenuated summer flow recession in the Little Sandy River compared to Johnson Creek. Average precipitation in the Little Sandy basin is 106 in./yr compared to 59 in./yr for the Johnson Creek basin at the Sycamore site, based on PRISM data from 1971 to 2000 (PRISM Group, 2007). In all, these differences result in more than triple the flow per unit area of the Little Sandy River compared to Johnson Creek.

Annual Flow

The factors controlling streamflow of Johnson Creek on an annual time scale are the distribution of precipitation and groundwater discharge and routing of runoff from the natural and human-modified landscape. More rain falls in the southern and eastern areas of the basin, coincident with increases in elevation. Precipitation, once on the landscape, can follow several pathways, where it can be delivered rapidly to the stream, or where it may enter the groundwater flow system as recharge and eventually discharge to the stream or out of the basin. Alterations in the form of roads, ditches, drywells, municipal sewer systems, and agriculture have an effect on the distribution of runoff; however, many of these changes may have predated collection of streamflow data in the Johnson Creek basin.

Spatial Trends

Spatial trends in annual flow were analyzed using streamflow and precipitation data from WY 1999 to 2006. The streamflow data used in this analysis are from Johnson Creek at Gresham (map number 26), Sycamore (in Portland) (map number 30), and Milwaukie (map number 41), and from measured flows of Crystal Springs Creek (map number 40). Comparisons were made of the flow at each site, of the inferred inflow between sites, and of flows compared to precipitation. Streamflow per unit area is greatest in the eastern area of the basin upstream of the Gresham site. This pattern of runoff is a result of three factors: (1) precipitation is greater in the upland areas because of the topography; (2) the less permeable, fine-grained soils and the relatively steep topography of the upland areas favor rapid runoff to the stream; (3) part of precipitation enters the combined sewer system in the western, more urban areas of the basin. Although Johnson Creek downstream of the Gresham site receives less runoff than would be expected in per unit area, focused groundwater discharge in some areas yields relatively high base flow. This high base flow is evident in Crystal Springs Creek, which features large flows compared to its relatively small surface-water contributing area.

Although spatial comparison of streamflow in per-unit-area generally is useful, this measurement poses challenges because of the coupled nature of the groundwater and surface-water flow systems, and human causes. Streamflow at a given location on Johnson Creek is the result of surface runoff and the net discharge from the groundwater flow system. The surface-water contributing area is a relatively static feature based on the topography. In an urban setting, the flow divides defining the contributing area can be altered by the configuration of street drains and other development. The groundwater contributing area, although it loosely mimics the topography, also is affected by groundwater flow direction (which varies with depth), and the hydraulic head in aquifers. Human-caused alterations additionally are in the “plumbing” of the drainage basin, such as expedited delivery of precipitation to the stream in agricultural areas by tile drains (reducing potential groundwater recharge), diversion of storm runoff to drywells in urban areas (reducing surface runoff to the stream), and diversion of storm runoff to the combined sewer system that discharges out of the basin.

Annual mean streamflow increased from the Gresham to the Sycamore to the Milwaukie sites based on streamflow records from WY 1999 to 2006. The average streamflow of Johnson Creek from Gresham to Sycamore increased by 54 percent (from 28.7 to 44.2 ft³/s), and an additional 53 percent (to 67.8 ft³/s) from Sycamore to Milwaukie. About one-half of the increase in streamflow between the Sycamore and Milwaukie sites is from Crystal Springs Creek.

The streamflow in the three areas of the Johnson Creek basin was represented in per-unit-area. Streamflow in the upper basin is represented by the Gresham site. Streamflow in the middle basin is represented by the difference in streamflow between the Gresham and Sycamore sites. The lower basin is represented by the difference in streamflow between the Sycamore and Milwaukie sites. Dividing the average streamflow by the drainage basin area to derive a streamflow per-unit-area value enables comparison of different size areas of the drainage basin, and expression of streamflow per unit area in inches enables comparison to precipitation. The upper basin contributes more streamflow to Johnson Creek in per unit area than does either the middle or lower basins. The average streamflow (WY 1999–2006) of the upper basin was 25.3 in. compared to 18.5 and 12.1 in. in the middle and lower basin, respectively.

Estimates were made of precipitation falling on each area of the Johnson Creek basin to relate precipitation to streamflow in the area. Precipitation was estimated by determining the ratio between the PRISM-derived precipitation at the Portland Airport and of the upper, middle, and lower areas of the basin (PRISM Group, 2007). These ratios were applied to the measured precipitation at the Portland Airport from WY 1999 to 2006 (Oregon Climate Service, 2007). Streamflow as a fraction of precipitation decreases from 45 percent in the upper basin to 37 and 28 percent in the middle and lower basin, respectively.

The differences in the delivery of precipitation to the stream are a result of geomorphic settings and human causes. The upper area of the basin consists of moderate to steep terrain and fine sediments. These factors favor runoff from the land surface. A dense network of roads, ditches, and farm fields generally direct runoff away from the land surface and toward tributaries or the mainstem Johnson Creek. Compared to the upper basin, the lower basin generally is lower relief. Stream channel development, particularly on the north side of the basin is minimal because of the low relief and coarse-grained permeable Missoula Flood deposits. The present and predevelopment landscape feature minimal stream channel development. This area was shaped by deposits from catastrophic Columbia River floods, leaving a relatively flat and highly permeable landscape. Maps made by the General Land Office between 1852 and 1855 provided detailed descriptions of natural-resource features, and indicated few streams in the northern area of the basin (University of Oregon, 2006). Although the lower area of the Johnson Creek basin primarily is a dense urban grid, runoff from streets, rooftops, and other impervious surfaces to the creek is curtailed through interception by the combined sewer system and drywells. The middle basin integrates features of the upper and lower basins, representing a mix of upper-basin and lower-basin characteristics, resulting in moderated streamflow relative to precipitation.

Temporal Trends

Temporal trends in annual flow in the Johnson Creek basin were identified through analysis of streamflow and precipitation data. The 66 years of record (WY 1941–2006) at the Sycamore site (map number 30) indicate long-term trends that are a response primarily to variation in precipitation. The cumulative departure from annual mean values is shown in fig. 16 for the periods of record of streamflow of Johnson Creek at Sycamore and at Milwaukie (map number 41), and of annual total precipitation at the Portland Airport. Annual mean streamflow of Johnson Creek at Sycamore, from WY 1941 to 2006 ranged from 15.6 ft³/s (WY 1977) to 91.7 ft³/s (WY 1997). Decade-scale dry cycles (such as from 1956 to 1967 and from 1984 to 1994) were accompanied by decreased annual mean streamflow. The relatively short period of record for Johnson Creek at Milwaukie tracks the end of a dry cycle in about WY 1994, follows the relatively wet years of WY 1995–1999, and again includes a drying trend from WY 2000 to 2005.

Temporal trends were tested in streamflow of Johnson Creek at Sycamore as a fraction of precipitation from WY 1941 to 2006 using the Kendall’s tau correlation statistic. In this way, although precipitation varied over time, trends in the delivery of that precipitation to the stream on an annual scale were assessed. The null hypothesis is that the ratio of annual streamflow to precipitation is constant over time. A plot of the ratio of annual mean streamflow to precipitation over time indicated no visual trend and was substantiated by the Kendall tau value of -0.083 (a slight decreasing trend), and a p-value of 0.32, indicating that this trend was not statistically significant. The generally close relation of annual mean streamflow of Johnson Creek at Sycamore from WY 1941 to 2006 compared to annual total precipitation indicates that little has changed on an annual scale in the response of streamflow to precipitation.

Low Flow

Analyses of low streamflow, in terms of spatial and temporal variability, are needed to understand the effects of past land-use practices and the potential effects of future activities. Low streamflow and warm stream temperature has the potential to negatively impact the beneficial uses of the stream by wildlife and people.

Factors that control the low-flow regime of Johnson Creek include the general topographic and geologic setting, land use, groundwater flow direction and discharge, climate, and water use. The low flow period typically is in the late summer. In this report, discussion of the groundwater flow system preceded discussion of streamflow, and in particular, low flow, because groundwater is the primary source of streamflow during the dry summer period. An understanding of low flow was derived in part from analysis of seepage measurements and low-flow characteristics at the streamflow sites on Johnson Creek. This analysis adds to the previous discussion, including analysis of streamflow data on Johnson Creek since 1940, and possible human-induced changes that may have exacerbated low flows in some years. The data used for this analysis are daily mean streamflow of Johnson Creek and Little Sandy River, streamflow measurements, and precipitation.

Spatial Trends

Low flow varies considerably across the Johnson Creek basin because of differences in groundwater discharge. Although precipitation is the initial source of streamflow, rainfall during the summer is minimal. The minimal rain in summer mostly is intercepted by the dry soils. Because of relatively high temperatures and active plant growth, water on the land surface and the vegetated canopy rapidly evaporates or transpires. Flow-duration curves were used in the analysis of spatial trends in low flow. High flows are exceeded only a small fraction of the time, whereas low flows are exceeded most of the time. Differences in the lower end of the flow-duration curve indicate differences in flow characteristics related to the presence or absence of groundwater discharge to the stream.

Flow-duration analyses were based on daily mean streamflow from WY 1999 to 2006 at the streamflow sites in Gresham, Sycamore, and Milwaukie (map numbers 26, 30, and 41). The flow-duration curves representing the intervening areas were determined by subtracting the daily mean flow at the upstream site from that of the downstream site, which yielded the daily mean contribution of flow from the upper, middle, and lower areas of the basin. The flow associated with the percentage of time that was equaled or exceeded was divided by the size of the drainage basin area, which resulted in flow duration per square mile. Although the curves represent the range of flow duration, the focus of this analysis is the low-flow part (right side) of the flow-duration curve when a given flow is exceeded most of the time during the extreme low flow period.

The flow-duration curves shown in figure 17 illustrate the low flow characteristics in the three areas of the Johnson Creek basin. Notable characteristics of the low-flow end of the flow duration curves are the relative similarity of the curves for upper and middle areas of the basin and the distinct contrast with the curve for the lower area of the basin. The low-flow segment of the flow duration curves for the upper and middle areas of the basin are typical of streams that receive little groundwater discharge and that have rapidly decreasing streamflow during seasons of low precipitation. The lower basin, in contrast, features a sustained flow, particularly at flow duration greater than 60 percent, which is a result of groundwater discharge through springs in the lower area of the Johnson Creek basin. Although the broad, relatively low-relief northern area of the basin downstream of Sycamore contributes minimal flow, spring and seepage discharge to Johnson Creek downstream of RM 5.5 sustains summer flows. For example, at 90 percent flow duration, although the streamflow of the upper and middle areas of the basin is about 0.07 (ft³/s)/mi², streamflow in the lower area of the basin is 0.4 (ft³/s)/mi².

Temporal Trends

Temporal treads in low flow in the Johnson Creek basin from WY 1941 to 2006 were determined from analysis of daily mean streamflow data. These analyses include comparison of annual 7-day and 30-day minimum streamflow each year, comparison of flow-duration characteristics and base-flow separation. Although summer and annual precipitation have an effect on minimum flows each year, water use may have a greater effect during some years. A water-use assessment was not part of the study, so the effect of water use was inferred from the absence of correlation between low flow of Johnson Creek and low flow of Little Sandy River.

Annual 7-day and 30-day minimum streamflow statistics were computed for sites on Johnson Creek and on Little Sandy River. Most of the low flows of Johnson Creek extended past the water-year boundary (September 30 of each year), so the period of analysis for annual minimum flow was the low-flow season from May through October of each calendar year. The annual 7-day minimum flow of Johnson Creek at Sycamore (map number 30), from 1941 to 2006 ranged from about 0.1 to 2 ft³/s, and the annual 30-day minimum flow ranged from about 0.2 to 3 ft³/s (fig. 18). Sustained high annual precipitation, characterized by the 3-year moving average precipitation from WY 1997 to 1999 (fig. 5) and resulting groundwater discharge to Johnson Creek, was the apparent cause of consistently elevated minimum flows during those years. Other years of particularly high (or low) 7-day or 30-day minimum flows did not appear to be as closely related to extremes in annual precipitation.

During the period from 1955 to 1977, the annual 7-day and 30-day minimum flows of Johnson Creek at Sycamore each year were, on average, about one-half the long-term mean values for annual 7-day and 30-day minimum flows. These sustained low flows occurred despite some large values of annual precipitation (1956 and 1974) and a general increasing trend in annual total precipitation from about 1948 through 1956 and again from about 1969 through 1976 (fig. 5). Annual 7-day and 30-day low flows of Little Sandy River (map number 45) did not indicate decreased summer flows during this period and actually were slightly greater compared to the period from 1941 to 2006. A potential cause for the particularly low streamflow of Johnson Creek during the 1955 through 1977 period is increased water use either from instream pumping or groundwater withdrawal that captured groundwater that would otherwise have discharged to the creek. The close relation of the 7-day and 30-day minimum flows shows a long-lasting decrease in summer flows of Johnson Creek during this period. After 1977, the 7-day and 30-day low flows generally increased in Johnson Creek indicating perhaps that the period of increased water use had ended.

Temporal trends in low flows also were assessed using flow-duration analyses. The period of record of Johnson Creek at Sycamore (map number 30) was divided into three segments of 22 years each, thus providing flow duration for the WY 1941–1962 (early), WY 1963–1984 (middle), and WY 1985–2006 (later) periods (fig. 19). The result of the flow duration analysis is consistent with that of annual minimum flows. The lower part (right side, low flow) of each duration curve is the focus for this analysis. Low flows, for example at 90 percent flow duration, during the early period were greater than in the middle period. The cause for this difference in low flows is likely the extreme low flows in the summers of 1955–77 (representing much of the middle period) as identified in the annual 7-day and 30-day minimum flow analyses (fig. 18). The later period indicates the greatest low flows of the three periods analyzed. A comparison of the extremes represented by the middle period and later period flow-duration curves shows that daily mean flow 90 percent of the time has increased from 0.9 to 1.7 ft³/s.

The base-flow component of streamflow may be affected by changes in the drainage basin over time. Changes typically associated with urbanization may reduce base flow because of reduced infiltration and diversion of storm runoff into the sewer system. The base-flow component of streamflow for Johnson Creek at Sycamore for the period from 1941 to 2006 was determined using the program PART (Rutledge, 1998). Base flow ranged from about 45 to 70 percent of total flow. A trend test using Kendall’s tau indicated a slight decreasing trend in percent base flow (a tau of -0.14); however, the p-value was 0.09, indicating that the trend was not significant at the 0.05 level. The test was done on Little Sandy River for the same time period and also indicated no trend.

Trends identified in the low flow of Johnson Creek may relate to water use but are not necessarily associated with land-use changes characteristic of urbanization. The cumulative impact of urban development upstream from the Sycamore site from 1941 to 2006 has not affected the magnitude of low flows based on the overall absence of trend observed in the 7-day and 30-day low flow from 1941 to 2006, and the base-flow component of streamflow. This does not mean that the present low-flow condition is the same as prior to development in the basin, because much of the infrastructure was already in place prior to the beginning of streamflow data collection. These data do indicate, however, that changes in land use may have had less effect on low flow than water use, which may have had an increase around 1955 and subsequently decreased around 1977. As land uses change, primarily from agricultural and natural areas to a dense urban grid, particularly in upland areas, continued streamflow data collection and analyses are needed to document hydrologic response in the low-flow regime of Johnson Creek.

High Flow

Flooding of Johnson Creek likely was a concern when people began to inhabit areas near the creek. Prevention of flood damage was the primary incentive for the large-scale channel modification project that resulted in the present configuration of the lower 11 mi of the creek. Johnson Creek floods as a result of intense rainfall, usually over a several day period, and sometimes melting snow adds to the flood. In contrast to annual and low flow on Johnson Creek, where the magnitude of flow is in part controlled by the groundwater system, the presence of high flow mostly is a response to runoff from the land surface. The degree of flooding typically is determined by precipitation intensity and antecedent soil moisture. Other factors that can contribute to spatial and temporal variability in high flow events are temperature, topography, soil type, vegetation, channel configuration, the network of roads, ditches, storm drains, sewer systems, and impervious surfaces. Changes in the frequency of flooding, particularly related to changes in land uses in the basin have been a topic of discussion for decades. The record of streamflow of Johnson Creek at Sycamore (map number 30) are relatively long, dating back to WY 1941. The Johnson Creek streamflow record at Sycamore and the shorter-term records at Milwaukie (map number 41; since WY 1990) and at Gresham (map number 26; since WY 1999) provide data for comparison and assessment of temporal trends in high flow across the basin.

Spatial and temporal trends in high flow of Johnson Creek were evaluated based on annual peak flow and daily mean flow. Stream level during high-flow conditions is often of primary interest. Flow is computed at each streamflow site and the stream level is related to a peak streamflow using a stage-discharge rating. Changes as a result of scour and fill in the stream channel, and natural and constructed features in and near the stream can affect the level at a given streamflow. Flood-frequency statistics were computed for the Johnson Creek at Sycamore and at Milwaukie sites. Daily mean streamflows were used for flow duration analysis at the Gresham, Sycamore, and Milwaukie sites, and comparisons of the fraction of days in a year that are greater than the annual mean streamflow were made at the Sycamore and Milwaukie sites.

Spatial Trends

Analyses of spatial trends in high flow in the Johnson Creek basin include comparison of peak flows and volumes, flood frequency, flow duration, and the fraction of daily mean flows that are greater than the annual mean streamflow. Better understanding derived from comparison of peak flows is compared to previous studies of the area.

The annual peak streamflow at the Milwaukie site can be either greater or less than at Sycamore, contrary to what may be expected because of a two-fold increase in basin area (table 3, fig. 20). The annual peak usually is on the same day, and only in WY 2001 (an extremely dry year) was the peak of the year at each site the result of a different storm. The difference in annual peak streamflow varied widely. The annual peak streamflow at the Milwaukie site was between 24 percent less (WY 1997) and 53 percent more (WY 2005) than at the Sycamore site. The average increase in peak flow from the Sycamore site to the Milwaukie site, based on peaks from WY 1990 to 2006, is 11 percent. The difference in peaks after WY 1997 may have increased (table 3); however, the record of concurrent peaks is lacking. Uncertainties in the stage-discharge rating at the Milwaukie site (because of the relatively short period of record compared to the Sycamore site) additionally may affect the computation of peak flow.

The peak flow at each site is a function of the spatial and temporal distribution of precipitation and runoff characteristics of each area of the Johnson Creek basin. The timing and magnitude of precipitation (and subsequent runoff) in the area of the basin between the Sycamore and Milwaukie sites is not the same as for the area of the basin upstream of the Sycamore site because of the general east-west orientation of the basin; the tracking of storms, most of which is from west to east; and the generally greater precipitation in the areas of the basin at higher elevations. Depending on rainfall patterns and intensity, the response in flow at the Milwaukie site may be a single (and potentially larger) peak, an attenuated response, or peaks separated by several hours. Peaks at the Milwaukie site additionally may be attenuated by interception of precipitation in the relatively flat, permeable deposits on the northern side of the basin, and by interception by drywells and combined sewer systems.

The relation of peak streamflow to stream level is fairly consistent at Johnson Creek at Milwaukie; however, the level of the Willamette River in the Portland area can cause backwater conditions in the lower area of Johnson Creek, extending an undetermined distance upstream of the Milwaukie site. Backwater from the Willamette River seems to affect Johnson Creek at the Milwaukie site when the level of the Willamette River exceeds about 22 ft, which is recorded twice after records at the Milwaukie site began in WY 1990. On February 8, 1996, at the time of the peak level at the Milwaukie site, the level of the Willamette River at Portland (map number 44) was about 22 ft. Although the flow of Johnson Creek decreased following the peak, because of the relatively large size of the Willamette River basin, the Willamette River at Portland continued to rise, peaking at about 27.7 ft on February 9. This rise of the Willamette River resulted in a rise in stream level of Johnson Creek of about 4 ft, despite decreasing streamflow of Johnson Creek. The backwater condition at the Milwaukie site lasted for several days resulting in the flooding of nearby properties. More backwater at Johnson Creek at Milwaukie was recorded in January 1997 when the Willamette River at Portland peaked at about 23 ft causing backwater at the Milwaukie site for about 2 days.

Although peak volume increases between the Sycamore and Milwaukie sites, the increases are not proportional to the increase in the size of the drainage basin area between the two sites. The peak volume each year was defined as the cumulative daily mean streamflow during the 6-day period surrounding the annual peak (table 3). The 6-day period was selected by visual inspection of the instantaneous streamflow hydrograph, beginning the day of increase in flow and extending several days past the day of peak flow. Antecedent flow was not taken into account in this analysis because it was relatively low (compared to the peak flow) and fairly similar at the two sites. The average increase in annual peak volume between the Sycamore and Milwaukie sites from 1990 to 2006 was 24 percent, and ranged from 7 to 40 percent. Factors contributing to the relatively small increase in runoff volume between the Sycamore and Milwaukie sites may be the increased infiltration through the permeable surficial deposits, capture of runoff from the urban area between the two sites, and subsequent transport of storm runoff either to drywells or to the combined sewer system.

Flood frequency statistics for Johnson Creek at Milwaukie based on annual peaks from 1990 to 2006 were compared to those for Johnson Creek at Sycamore for the same period (fig. 21, table 4). This analysis indicates that streamflows for recurrence intervals greater than about 10 years are greater at the Sycamore site than at the downstream site. The smaller values at the Milwaukie site may indicate that peak flows at the upstream site that are in excess of the channel capacity are temporarily stored on the floodplain and do not contribute to the peak flows at the Milwaukie site. Flood-frequency statistics for Johnson Creek at the Sycamore site for the entire 1941–2006 period of record also are shown in figure 21 and table 4. Flood-frequency statistics generally are more reliable for longer periods of record, but the differences between the 1990–2006 period and the 1941–2006 period are slight for this site.

Similar to analyses of low-flow duration statistics, the high-flow part of the duration curve indicates differences in flow-duration characteristics of the upper, middle, and lower areas of the basin. At a flow duration of 5 percent (where daily mean streamflow is equaled or exceeded 5 percent of the time), as shown in figure 17, the area of the drainage basin upstream of the Gresham site contributes 8 (ft³/s)/mi², compared to 5 (ft³/s)/mi² between the Gresham and Sycamore sites and 2 (ft³/s)/mi² between the Sycamore and Milwaukie sites. When the contribution to high flow of the upper, middle, and lower areas of the basin are compared, differences generally are a result of decreased precipitation, gentler slopes, more permeable soils, and greater interception by drywells and combined sewer systems toward the west.

An assessment based on daily mean streamflow at the Sycamore and Milwaukie sites from 1990 to 2006 was made of the “flashiness” of the stream, which is the tendency of the streamflow to rise from a base-flow condition rapidly and decline rapidly. The indicator statistic is the percentage of days in a year where the daily mean streamflow is greater than the mean streamflow in that year. In a basin with rapid, short-duration, or “flashy” runoff, a smaller percentage of days exceed the mean streamflow for the year when compared to a basin in which streamflow rises and declines more slowly. Konrad and Booth (2002) and Chang (2006) showed that this statistic was an indicator of urbanization, affecting streams in Washington and in the Portland, Oregon, area (including Johnson Creek). The consequence of rapid changes in flow on a stream may be degradation of the stream channel, erosion, and disruption of biota in the stream. The mean flow of Johnson Creek at the Sycamore and Milwaukie sites was exceeded an average of 29 percent of the time. The same analysis was made of Little Sandy River (map number 45), where the mean streamflow was exceeded an average of 34 percent during the same period. These results are consistent with previous studies that showed the mean streamflow was exceeded a smaller fraction of the time in urban streams than in rural streams (Konrad and Booth, 2002). The similarity of this statistic at the Sycamore and Milwaukie sites primarily is because of the absence of storm-induced rapid runoff in the lower area of the basin.

The relatively small change in flood-frequency of Johnson Creek at the Sycamore and Milwaukie sites, especially the decrease in flood flows for recurrence intervals greater than about 10 years, is not consistent with predictions of flood frequency from methods currently used in Oregon. Although the available flood prediction equations developed for use in Oregon (Cooper, 2005) are applicable only to rural streams in western Oregon and thus may not be reliable for urbanized basins such as the Johnson Creek basin, equations for use on urbanized streams in the Portland area were developed by Laenen (1980). Laenen’s prediction equations, which included data from the Sycamore site on Johnson Creek, used basin drainage area and a basin-storage term as explanatory variables. The form of the prediction equations indicated that flood flows for a given recurrence interval increase with increasing size of the drainage basin area, but that flood flows for a given recurrence interval decrease with increasing basin storage. Results from the prediction equations developed by Laenen (1980) are particularly sensitive to estimated values of basin storage and basin storage is difficult to determine and perhaps not directly related to factors that may be affecting flood runoff in the lower area of the basin, such as increased storm runoff interception. Based on the available data and definitions provided by Laenen (1980), basin storage at the downstream Milwaukie site was not much different than basin storage at the upstream Sycamore location. Application of the prediction equations, using equal values for basin storage but considerably larger drainage basin size at Milwaukie than at Sycamore, yielded peak-flow estimates for the Milwaukie site that were about twice as large as those from the log Pearson 3 analysis of the measured peak streamflow. As a result, the Laenen (1980) equations do not reliably account for the factors that affect flood-frequency statistics on Johnson Creek, although they may be reliable for prediction purposes on other urbanized streams in the Portland area.

Temporal Trends

Streamflow and water level at the Sycamore site (map number 30) were used in the analyses of temporal trends in high flow of Johnson Creek. High flows were characterized in terms of annual peaks, flow duration, and the fraction of daily mean flows that are greater than the annual mean streamflow.

A Kendall’s tau test applied to the 66-year record of annual peak flows (fig. 20) indicated an absence of a temporal trend in annual peaks; the value of tau was -0.03 and the p-value was 0.70. Although the basin has changed much between 1941 and 2006, those changes have not led to a trend toward higher or lower peak flows.

Because a concern during flooding is the stream level rather than streamflow, an analysis was made of the trend in annual peak stream level associated with each annual peak streamflow at the Sycamore site from WY 1941 to 2006. The stage-discharge rating curve in use in WY 2006 is shown in figure 22; points represent the annual peak stream level andstreamflow in the early (WY 1941–62), middle (WY 1963–84), and later (WY 1985–2006) periods. Most peaks from the later period plot along this curve. Many peaks in the early and middle periods plot to the right of the curve (greater streamflow for a given stream level), indicating, in general, that the capacity of the stream channel to carry high flow was greater in the early and middle periods than in the later period.

The change in the relation of stream level to streamflow is from about 700 to 1,500 ft³/s, and amounts to a 1-ft aggradation (filling) of the stream channel. At streamflow less than about 700 ft³/s, the peak stream level relative to streamflow is more variable than at higher flow, which is a result of the impermanence of vegetation, large wood, and other objects in the stream channel. Streamflow greater than about 1,500 ft³/s exceeds the capacity of the trapezoidal channel and spills into the overbank area. At this stream level, the stage-discharge rating changes little over time because of the general consistency of the shape and slope of the terrain near the streamflow site. This analysis shows that the stream channel gradually was filled in, but relative to the range of stream level (about 15 ft) and the time period (66 years), the 1 ft of fill is small. When flows are greater than about 2,000 ft³/s, corresponding to a recurrence interval of 10 years and more, the relation of stream level to streamflow has not changed measurably. This analysis is valid for the Sycamore site only. Other changes to the stream channel elsewhere in the basin may have altered the relation of stream level to streamflow.

The capacity of a stream channel can change naturally over time and may or may not be a result of human activity. Aside from the initial construction of the stream channel in the 1930s (prior to the beginning of streamflow record at the Sycamore site), the largest single change seems to be in 1968 when much of the channel of Johnson Creek was cleared of brush and debris. County work crews cleared the channel in August of that year and extensive clearing along the creek was reported in an article (Johnson Creek Watershed Council, 2006) citing memories of a once young man employed along with “hundreds of Portland boys for the summer” (of 1968), using “two-man saws, axes, brush hooks, etc. … in an effort to stop the annual flooding.” The result of this clearing lasted for several years afterward; now the channel has refilled from an accumulation of vegetation and debris.

Flow duration was used to evaluate trends in high flow of Johnson Creek and indicated little change from WY 1941 to 2006 in the percentage of time a given daily mean flow was exceeded. The period of daily flow record was divided into three equal-length periods: WY 1941–62 (early), WY 1963–84 (middle), and WY 1985–2006 (later) (fig. 19). Flow at a flow duration of 5 percent differed by less than 20 percent between the early, middle, and later periods.

The measure of “flashiness” of the stream was assessed over time by comparing the percentage of days in a given year that the daily mean streamflow was greater than the annual mean streamflow for that year. The period of record at the Sycamore site was again divided into the same early, middle, and later periods defined previously. The annual mean streamflow in the early and later periods was exceeded 29 percent of the time, but the annual mean streamflow in the middle period was exceeded 26 percent of the time. Temporal trends in “flashiness” were assessed using Kendall’s tau correlation coefficient. For the period from WY 1941 to 2006, the Kendall tau value was 0.04; however, the trend was not statistically significant as the p-value was 0.65.

The assessment of “flashiness” of Johnson Creek used daily mean (in contrast to instantaneous) streamflow. Comparison of the rate of rise and decline during storms using instantaneous streamflow may produce inaccurate results because of changes in stream-level monitoring instruments at Johnson Creek at Sycamore. From 1940 to 1997, the stream-level sensor consisted of a stilling well with a float-driven recorder (Rantz and others, 1982, p. 50). Because the stream-level recorder was inundated during high flows many times and the pipe (actually an 8-in. handmade stone culvert) that relayed the stream level to the stilling well occasionally clogged with silt, the hydrograph recorded a delayed response on the rising and falling limb of the hydrograph. The condition of the pipe could only be detected during periodic visits, so many delayed responses may have gone unnoticed. As a result, although the daily mean and peak streamflow may be accurate, the rate of rise and decline of the stream level may not be. In December 1997, the stilling well was replaced with a pressure sensor (Rantz and others, 1982, p. 52) that was much less subject to fouling and tracked the fluctuations in stream level more accurately. Comparison of the potentially erroneous delayed-response hydrograph from earlier times to a later response may lead to an incorrect conclusion about the rate of rise of the stream.