Skip Links

USGS - science for a changing world

Scientific Investigations Report 2010–5008

Use of Continuous Monitors and Autosamplers to Predict Unmeasured Water-Quality Constituents in Tributaries of the Tualatin River, Oregon

Introduction

Since the early 1990s, the quality of water and ecological health of tributaries to the Tualatin River in northwestern Oregon (fig. 1) have been the subject of heightened concern from resource managers, regulators, and citizen groups. The small urban and agricultural streams on the eastern side of the basin are known to have water-quality problems, but the magnitude, duration, seasonality, and short- and long-term trends for those concerns have not been well characterized. Aspects of those problems have been studied, including low-flow phosphorus and bacteria levels (McCarthy, 2000), storm-related variations in nutrient and bacteria concentrations (Anderson and Rounds, 2003), and the levels of trace metals and organochlorine pesticides in fish tissue and sediment (Bonn, 1999). Issues of high water temperature, excessive bacteria levels, high phosphorus concentrations, and low dissolved oxygen concentrations were cited as particular problems requiring attention in Tualatin River tributaries in the 2001 revision of the Total Maximum Daily Load (TMDL) regulations for the basin (Oregon Department of Environmental Quality, 2001). Increased monitoring and additional studies have helped to fill gaps in our understanding of the dynamics of water quality in these streams. The characterization of the short-term dynamics, long-term trends, and spatial variations of water quality in these systems, however, probably will require the use of new approaches.

The use of submersible instruments that simultaneously measure and log multiple water-quality parameters in situ is growing rapidly in the Pacific Northwest and nationally. Such instruments can collect data at regular intervals and for long periods without human intervention, thereby providing opportunities for increased data collection at reduced costs. These instruments often are referred to as continuous monitors because they can be operated continuously for long periods. Data from continuous monitors can be used for many purposes, including (1) documentation of routine or event-based environmental conditions in a drainage basin, (2) detection of daily and seasonal variations and long-term trends in water quality, (3) calibration and validation of numerical models, (4) feedback for regulatory and resource management systems, and (5) surrogate measurements for the calculation of concentrations or loads of suspended sediment (Gray and Glysson, 2003; Uhrich and Bragg, 2003) or other constituents (Christensen and others, 2000). Monitored parameters typically include water temperature, specific conductance, pH, and, increasingly, dissolved oxygen, turbidity, and chlorophyll. Many other types of sensors are under development.

Despite the advantages of these continuous monitors, many constituents of interest to regulators and resource managers still cannot be directly measured by such technology. For example, streams in the Pacific Northwest often are managed for their concentrations of suspended sediment (or total suspended solids), various nutrients (nitrogen and phosphorus species), or bacterial pathogens (such as Escherichia coli [E. coli] as an indicator of bacterial pathogens). No routine and direct in situ measurements can be done for these constituents at environmentally relevant concentrations by currently available commercial instruments. Such analyses, therefore, must be made in a laboratory using discreet samples collected from the stream.

Data from continuous monitors, however, sometimes can provide an indication of the concentrations of unmeasured constituents. For example, turbidity in water often is directly dependent on suspended sediment concentration (Lewis, 1996; Anderson and Rounds, 2003; Gray and Glysson, 2003; Uhrich and Bragg, 2003); therefore, turbidity data from continuous monitors can be used to estimate a time series of suspended sediment concentration. Christensen and others (2000) used data from continuous monitors in Kansas streams to calculate instantaneous concentrations and loads of alkalinity, dissolved solids, total suspended solids (TSS), chloride, sulfate, atrazine, and fecal coliform bacteria. Site-specific regressions between monitored parameters and the results of discrete water samples were derived for these constituents, and the regressions were then applied to long-term monitor records at the study sites to estimate a time series of constituent concentrations. By combining these concentration estimates with discharge information, constituent loads also can be estimated. For example, Uhrich and Bragg (2003), Anderson (2007), and Bragg and others (2007) performed similar calculations using continuous records of turbidity and discharge to estimate suspended sediment concentrations and loads in the North Santiam and McKenzie Rivers, respectively, in western Oregon.

To use continuous monitors to develop robust statistical models for sampled water-quality constituents, independent samples representing a broad range of conditions (high- and low-flow and seasonal warm/cold or spring/summer/autumn/winter) are needed at each site. Clean Water Services, the primary wastewater and stormwater management utility in the urban areas of Washington County, Oregon, has been collecting routine water-quality samples at many sites in the Tualatin River basin for more than 20 years. Most samples collected, however, represented low- or base-flow conditions, and were not targeted for storms. The U.S. Geological Survey (USGS) has collected data for many years and for various purposes at Fanno Creek near Durham, including during a few storms, but these data also are of limited scope. Data from these two sources were used to evaluate the potential regression models for this study.

Like continuous monitors, automatic samplers (referred to as autosamplers in this report) can collect water samples at night, during storms, or at specific intervals without the need for human operators. An autosampler can collect multiple samples (typically as many as 24) before it must be restocked with empty bottles. The autosampler also can be refrigerated or stocked with ice to minimize sample degradation. After collection, samples from the autosamplers (or autosamples) are retrieved and analyzed at a laboratory for the water-quality constituents of interest. Autosamplers can be programmed to collect samples at prescribed intervals of time or flow, and can be triggered by specific conditions. Used together, a continuous monitor can trigger an autosampler during an event (for example, when conditions exceed some threshold measured by the monitor) and can thereby document water‑quality conditions in the stream at the time of sample collection.

Continuous monitors and autosamplers offer many advantages over manual sampling, including the potential to collect many samples and large amounts of data during a short time, when the number of sites is large, if the sites are remote, or if the sites are difficult or inconvenient to access (such as at night, on weekends, or under hazardous conditions). These advantages are particularly useful when trying to characterize short-term variations in stream conditions during storms at multiple sites. Collecting an adequate number of samples at multiple sites during a storm with a crew of technicians can be inefficient and expensive compared to the use of remote and automated instruments, if they can accomplish the same tasks.

Despite these advantages, continuous monitors and autosamplers are subject to mechanical malfunction, sampling bias, or both, and require a certain degree of quality control to assure that the resulting data are accurate and representative of stream conditions. The quality control issues include the degree to which measurements or samples collected at one location in the stream by the autosampler represent conditions throughout the stream cross section, measurement bias because of fouling or sensor drift of deployed monitors, the possibility of carryover contamination because autosampler tubing was not completely cleaned, and the potential for exceeding prescribed sample holding times or temperatures in autosamplers. These issues must be addressed to ensure proper use of this technology.

This report uses correlative techniques that have been shown to work with relative success in various geologic regions (Christensen and others, 2000; Lietz and Debiak, 2005; Rasmussen and others, 2008), although the rivers studied typically have been larger than the Tualatin River tributaries. Application of these techniques was attempted in small Pacific Northwest streams that have large changes in characteristics between low and high flow, and in agricultural and urban areas. As part of a long-term scientific collaboration between the USGS and Clean Water Services, this study evaluated the quantity and attributes of data that are necessary to build useful predictive models for such streams.

Purpose and Scope

The purpose of this report is to evaluate the use of continuous monitors and autosamplers to collect representative and accurate water samples over a range of stream conditions, and to construct and demonstrate the use of preliminary predictive statistical models of unmeasured water-quality constituents in selected tributaries to the Tualatin River. Specifically, the objectives were to

  1. Evaluate the use of autosamplers for unattended sampling in conjunction with continuous monitors and evaluate the quality of autosamples;
  2. Develop preliminary regression models to predict the concentrations of selected water-quality constituents using concurrent data from continuous monitors, and evaluate the robustness and accuracy of those models;
  3. Evaluate the adequacy of available laboratory data to augment autosampler-derived data for developing regression models that estimate constituent concentrations and loads;
  4. Use the regression models to predict and evaluate time series concentrations and to develop uncertainty estimates for modeled constituent concentrations from historical continuous monitor data at the same sites; and
  5. Identify potential changes to sampling strategies that would allow future monitoring efforts to improve the regression models developed.

Sites on six selected tributaries in the Tualatin River basin were studied from June 2002 through December 2003. The sites represented a range of upstream land uses, from intense urban development to rural agricultural and forested areas. Continuous, in situ monitors recorded stage, streamflow, water temperature, specific conductance, dissolved oxygen, pH, and turbidity. Autosamples were analyzed for a suite of nutrients (nitrogen and phosphorus species), total suspended solids, chloride, and bacteria (E. coli) over a range of stream conditions. Approximately 48 discrete autosamples were collected at each site over the course of two or three storm events. The model-building process was augmented by additional data from USGS and Clean Water Services databases, covering the study period 2002–07.

Full development of example models was limited to two target sites—Fanno Creek near Durham Road, and Dairy Creek at Highway 8 near Hillsboro. Data from USGS and Clean Water Services databases were used to augment the autosampler data. The additional USGS and Clean Water Services data were concurrent with the dates of monitor deployment and, together with the continuous monitor data, were used to evaluate calibration and validation scenarios for these sites. For the remaining four non-target sites, preliminary model forms were identified but no additional data exploration was performed. Data were compiled for all samples collected using continuous monitors and autosamplers, but model development was limited to three whole-water constituents of primary interest to Clean Water Services and other local regulatory and resource-management agencies, specifically total suspended solids (TSS), total phosphorus (TP), and E. coli bacteria.

Study Area Description

The Tualatin River is a major tributary to the Willamette River near Portland in northwestern Oregon (fig. 1). The characteristics of the Tualatin River basin have been described in several reports, including those by Kelly and others (1999) and Rounds and Wood (2001). The basin has undergone rapid urbanization since the late 1980s and is now home to about half a million people, mainly in the central and eastern part of the basin and within the urban growth boundary of the Portland metropolitan area. Beyond the urban growth boundary, the fertile soil of the valley floor supports a wide variety of agricultural activities. The Coast Range Mountains to the west are densely forested and are a source for water supply and lumber production.

In an attempt to improve water quality in the Tualatin River and address specific issues related to algal growth and periodic high pH and low dissolved oxygen conditions, TMDLs for ammonia and phosphorus were set for the Tualatin River and its major tributaries in 1988 (Oregon Department of Environmental Quality, 1997), but primarily focused on the main stem Tualatin River. After the establishment of the TMDLs, studies of water quality in the main stem of the Tualatin River have highlighted the role of its tributaries as sources of TMDL constituents and oxygen-depleting substances to the main stem (Kelly, 1997). Results from the previous studies indicated that more information is needed on these constituents in the tributaries, and updated methods are needed to document their concentrations and delivery to the main stem during storm runoff periods.

When the Tualatin River TMDLs were revised in 2001, the tributaries received greater attention. New TMDLs for water temperature, bacteria, and oxygen-depleting substances were created, and modified limits on ammonia and phosphorus were retained (Oregon Department of Environmental Quality, 2001). Although the tributaries certainly affect the quality of water in the Tualatin River to some degree, the 2001 TMDLs demonstrated that the water quality and ecological health of the tributaries also was important.

Some river and tributary issues, such as high water temperature, algal growth, high pH, and low dissolved oxygen occur mainly during summer and autumn low-flow conditions, although high bacteria levels tend to be most problematic during storm events. An adequate characterization of the water-quality and ecological issues in the tributaries must include a good understanding of system behavior under a wide variety of conditions and time scales, and should address issues related to nutrients, bacteria, suspended solids, and other TMDL-related parameters. This study was designed to use continuous monitors to estimate some of these quantities, such as TP, TSS, and E. coli bacteria, and thereby aid in developing a better understanding of the dynamics of these parameters and how they affect stream quality.

Sites representing the broad range of land uses and hydrology in the basin were selected for development of regression models. The largest tributaries of the Tualatin River include Gales Creek (mainly forested), Dairy Creek (largely agricultural), Rock Creek (mixed urban), and Fanno Creek (urban). These creeks account for a large fraction of the drainage in the Tualatin River basin. A site on Beaverton Creek, a tributary to Rock Creek, has a large amount of upstream commercial and urban land use. Chicken Creek, a small tributary to the Tualatin River, was included because it drains a rapidly expanding urban and rural-residential area in the southern part of the basin. The locations and characteristics of these sites are shown in figure 2 and table 1. These creeks have water-quality problems that would benefit from further characterization.

First posted June 18, 2010

For additional information contact:
Director, Oregon Water Science Center
U.S. Geological Survey
2130 SW 5th Ave.
Portland, Oregon 97201
http://or.water.usgs.gov

Part or all of this report is presented in Portable Document Format (PDF); the latest version of Adobe Reader or similar software is required to view it. Download the latest version of Adobe Reader, free of charge.

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo USA.gov logo U.S. Department of the Interior | U.S. Geological Survey
URL: http:// pubsdata.usgs.gov /pubs/sir/2010/5008/section2.html
Page Contact Information: Contact USGS
Page Last Modified: Thursday, 10-Jan-2013 19:11:31 EST