Site Security and Access

The main consideration when evaluating a potential autosampler station is safety. Primarily this relates to the access and work area used by the field crew, but also relates to the physical safety of the structure and equipment. Wherever a site is accessible to the public, the station and equipment should be considered vulnerable to vandalism. Extra effort and expense will be needed to build a safe, secure structure. Structures should have a solid locking steel door and generally should not have windows. Depending on the location, it may be necessary to install power lines to a new station. Figure 2 shows a “hardened” pre-cast concrete sampling station in a public park equipped with a locking steel door and sited on a gravel pad. The structure was delivered to the site by the manufacturer. It is usually beneficial to locate the station at an existing secure site such as in a fenced-in area or within an existing building in a public park, for example. In these instances, a wood, metal, or fiberglass shed may suffice, saving considerable resources and time.

Legal Access and Permitting

Depending upon the location of a proposed station, it may be necessary to obtain Federal, State, county, or local permits. These may include stream encroachment permits, construction permits, and electrical permits if a power line is to be installed. Likewise, other legal agreements may be necessary to access land or limit liability. Obtaining such a permit may pose a significant delay in project work and should be considered when developing the project timeline.

Operating Conditions and Climate

Sampling stations may be subjected to extremes in operating temperatures. Sites that operate during freezing weather will need insulation and heaters for equipment, lines, and samples. Heat lamps or electric heaters can be used, but care must be taken to ensure proper electrical connections. Heat tape, which is often used to de-ice house gutters and water pipes, can be wrapped around the inlet line between the station and the stream channel and will usually necessitate access to alternating current (AC) line power. As the air temperature drops, power usage will increase and hasten the depletion of batteries. If subjected to hot temperatures, it may become necessary to cool sensitive electronic equipment using fans, vents, and air conditioning that require electrical service. As discussed later in this document, a critical concern in hot weather is sample preservation. Refrigerated autosampler units are optimal but may require AC power, or extra batteries that must be changed on an increased frequency, or if ice is used to preserve samples, then additional field trips will be required by crews.

Vulnerability to Flooding

Sampling sites are, by their nature, located along the banks of rivers, streams, and culverts that are subject to periodic flooding. Although rare, extreme floods can cause bank erosion and even destroy a station and its equipment. It is recommended that sampling equipment be located above the height of any likely flood stage; this stage may be estimated from the historic record of the streamgage (or nearby streamgages). However, if a peristaltic pump system is used, the height of a station must be below the limiting head of approximately 25 ft. Generally, peristaltic pump systems are most reliable if the pumping head and length of intake lines are minimized, which reduce pumping times, wear on pump tubing, and drainage of battery power. Careful consideration must also be given to the station foundation and the method used to secure the structure along the stream bank. Foundations are vulnerable to undercutting or destruction by flood water, and considerable effort is involved in pouring concrete footers and foundations for permanent stations; such steps often require following local building codes. For temporary stations, or sites not likely to be flooded, a suitable foundation may be created using crushed stone with shelters being simply staked down.

Utilizing an Existing Streamgage Station

In many instances, an existing streamgage station can be used to deploy the sampling equipment. If sufficient space is not available within the station, a semi-permanent shelter may be sited to house the autosampler equipment adjacent to the streamgage station; then the autosampler is connected to the existing electronic stage sensing and communication equipment and electrical service (if present). This add-on style of station can save considerable costs and time.

Deploying autosampler equipment in conjunction with an existing streamgage station also allows historic stage-discharge information to be accessed, which when combined with sampling data, allows constituent loads to be calculated. Stage data can be obtained either as an analog output from the existing pressure measuring system (digital stage recorder, pressure transducer, bubbler, sonic or radar system), or by obtaining digital data from the site’s data transmission system (radio, satellite, phone line).

Another advantage of deploying at or near an existing station is the access to the target stream. Additionally, most stations are located where cross-channel calibration sampling, either by wading or from a nearby bridge, can be safely conducted.

Shelters and Trailers

It is often possible to deploy temporary or semi-permanent shelters to house equipment short-term studies. There are numerous commercial or hand-built shelter designs available. Figure 3 shows examples of shelters that have been used by the USGS across the country; these include wooden, fiberglass, steel, or concrete structures. For temporary installations, inexpensive trailers can be deployed which are chained and locked to bridge railings and trees. Regardless of the installation, the shelter needs to sit on a solid foundation as close as possible to the stream but at an elevation above expected flooding level, and be secured with locks, chains, or cables.

Special Sampling Conditions

Autosamplers have been deployed to sample at many different non-riverine settings, these include:

Active Sampler Systems

Active autosampler systems consist of five general parts (figs. 5 and 6): (1) an intake line running between the autosampler and the sampled stream; (2) a pump to pull or push water through the intake line; (3) a source of electrical power (if applicable); (4) an electronic system that monitors environmental conditions, initiates and controls the autosampler, and records sampling data; and (5) a container and distributor system that allows one or more sampling containers to be filled. Commercially available active autosamplers are available that incorporate these five components, but in some instances, it may be necessary either to customize these autosamplers, or develop specialized autosamplers based on standalone components.

Passive Samplers

Passive samplers do not utilize electrically powered pumps to obtain water, but rather use chemical sequestration to remove constituents of concern from water bodies. This document is not intended to discuss passive sampling systems in detail; however, a brief discussion of this kind of sampler is available in Box 1 to help the reader understand the differences in information provided by passive and active sampling systems.

Pumping Devices

Commercially available autosamplers generally rely upon a peristaltic pump with internal, self-contained control, and a sample distribution and holding system. A typical autosampler is shown schematically in figure 7; this system includes a refrigerated chamber to hold sample containers. Alternatively, a component system can be developed around a peristaltic pump controlled by a stand-alone data logger. Usually, this system will produce only single bottle samples, either discrete or composited, although various more complicated systems utilizing valves can be designed to dispense to multiple containers.

An example of a novel autosampler used to collect multiple samples of stream water is shown in figure 8. This system uses a small submersible pump and valve system to collect multiple 40-milliliter (mL) volume samples for volatile organic compounds. The unit contains an internal microcontroller (or data logger), and the autosampler is lowered into a standpipe set into the riverbed (fig. 8B). A separate housing at the upper end of the standpipe contains communication equipment for remote monitoring. The controller turns the pump on at preselected time intervals and operates a valve system that selects the bottle to be filled. Water flows directly into the bottle through a needle; each sample requires no further handling or processing in the field and is sent directly to the analytic laboratory. Although this system was designed to collect samples on preset timed intervals, the programing system is adaptable to other triggering signals.

It is possible to modify commercially available autosamplers to fit specific project needs. For example, figure 9 shows an autosampler that was adapted to collect large volume samples (in this case, up to 75 liters [L] of water) required for analysis of trace organic constituents; in this instance the autosampler is housed in a small trailer to facilitate temporary deployments.

Advantages of the peristaltic pump systems include their commercial availability, their ability to be powered using batteries or line electricity, and their ability to reverse pumping direction, allowing the intake line to be rinsed and purged before each sample or aliquot is collected. Disadvantages include the limited height over which they can obtain water (generally less than 25 ft total head lift), a limited intake-line distance (due to head loss in the line), the need to replace worn or contaminated pump tubing, and the susceptibility of samples to degradation while stored in the sampler before retrieval.

The second type of active system uses a submersible pump that is installed in the streambed (figs. 10 and 11) and comes with a range of advantages and disadvantages (table 2). Submersible pumps push water directly to a sample container, or alternatively, into a tub from which a peristaltic autosampler can be used to retrieve a sample. These systems are generally (but not always) powered by electrical line service, with the submersible pump buried in a sump in the streambed or adjacent shoreline. The pump can be left to run continuously or by using a data logger or timer, set to operate only when a sample is needed. Like peristaltic systems, submersible systems rely upon separate electronic control equipment to initiate the sampling, control the volume of each sample or aliquot, and record sample information.

It is also sometimes advantageous to combine both submersible and peristaltic pumps to collect water samples. These systems generally fall into two categories where the submersible pump either (1) delivers water at pre-selected time intervals to a vat where it is available to a second peristaltic-based autosampler, or (2) continuously delivers river water through a line, from which a second autosampler withdraws water at preselected intervals. An example is shown in figure 12, where a data-logging control system is used to operate the submersible pump that periodically (15-minute intervals, concurrent with stage and discharge measurements at the streamgage) fills a plastic tub. Water-quality data sondes were set in the vat to measure a variety of parameters including dissolved nitrogen and various chlorophyll compounds. The inlet line from a peristaltic-pump based autosampler (not shown) could be secured into the tub allowing samples to be obtained from the vat immediately following each sonde measurement; an electric solenoid valve then automatically opens to drain the vat.

An example of another automated system was used by Bonin and Wilson (2006; modified from Litten, 1998) to sample river water for trace organic compounds in suspended sediment and water is shown schematically in figure 13. A submersible pump was used to continuously deliver water to the sampling station during storm events. The inlet of a peristaltic pump was attached to the submersible discharge line and used to periodically force a measured volume of water through a canister and flat filter, where suspended sediment was removed. A portion of the filtered water was then diverted using a positive displacement pump and was forced through a column containing XAD. Flow meters and water-weight were used to measure the volume of water processed; total volumes processed were typically 150 L through the filters, and 50 L through the XAD column. The sediment trapped on the filters for sediment-bound contaminants, and the XAD were eluted and analyzed for dissolved-phase constituents.

The third type of active system is the vacuum or siphon sampler; these use an evacuated sample container within the vacuum that causes water to be drawn directly into the container. An advantage of this system is cleanliness, as the water sample does not contact the pump. Disadvantages include the limitations of intake line length and height of water draw (typically less than 25 ft), vacuum leaks in the system may be difficult to eliminate, and these samplers cannot be used to collect water for analysis of dissolved gases or volatile organic compounds. An example of siphon sampler system is shown in figure 14.

Operating Power

All active autosamplers require electrical power and consideration must be given to power sources when designing and installing a system. At sites where the distance between the water source and the pump are short, or when deployments are temporary, 12-volt direct-current batteries can be used. Batteries may be provided by the autosampler supplier, or high-amperage automotive and marine batteries can be used. Typically, the electronic data-logging and telemetry equipment will use lower voltage-amperage batteries, solar power, or both. When batteries are used to power pumps, it is generally necessary to install freshly charged batteries immediately prior to a sampling event, and to connect several batteries in series as backup. If the autosamplers are prepared and ready to operate but are subjected to long intervals between uses or if extended pumping time is expected, then extra field effort must be planned.

Wherever possible, the use of AC-line power is preferred, especially when autosamplers are deployed for long periods of time and when submersible pumps are used. The use of heaters required for cold-weather deployments, or refrigeration during hot weather, will also generally require AC-line power. Line power will provide sufficient amperage, and removes the need to monitor and change batteries, and will certainly reduce the chance for missing samples due to battery failure. The possibility of unexpected voltage drops must be considered if long-runs of lines are needed between the service and remote pumps and equipment.

Finally, if line power is available or can be installed, the electrical service must be installed and checked by a certified electrician and must meet all applicable local and state electrical codes. It is imperative that special care is given to protecting field crews from electrical shock by using proper grounding and ground fault interrupter outlets by keeping floors dry and periodically inspecting the insulation of electrical lines.

Control Systems

Autosampler equipment is controlled using electronic systems that monitor time, stream, or other environmental parameters that determine the appropriate time to power the autosamplers to collect a sample, monitor specified volume of each sample or aliquot, record a date or time stamp for each sample, and log ancillary environmental data. The choice of using built-in control systems or stand-alone control systems to record, store, or transmit data should be made after careful consideration of project goals, budgetary constraints, distance from office to field sites, and other such factors.

Regardless of the autosampler system deployed, the system needs to record certain information for each sample or aliquot collected. At a minimum, a record of the date and time each sample or aliquot was collected. It is also advantageous to log additional instrument parameters such as number of pump rotations (per sample and total), battery power, pump pressure or suction, and other such mechanical and electrical variables. For example, monitoring the pump pressure and pump-head rotations per sample allows the condition of the pump tubing and intake line to be assessed. Table 3 presents some of the typical parameters to collect when operating an autosampler system.

Built-In Control Systems

Generally, the control systems of commercially available autosamplers are proprietary and specific to each manufacturer. It is imperative that the ease of operation, programming, and control adaptation are considered before new equipment is purchased. It is also imperative that the field crew be familiar with the programming and operation of the autosampler controls. Operation programs should be extensively tested in-house, and tested in the field, before project sampling is conducted. If difficulties are experienced in programing or selecting equipment, help can be obtained from the manufacturer (ISCO, 1994) or experienced researchers who can be found throughout the USGS.

Most commercially available autosamplers contain systems that allow researchers to monitor autosampler operations while on or offsite, although this often requires the purchase of additional, specialized equipment. It is important to consider the ease at which operation data can be accessed, and whether any ancillary data (such as that generated by water-quality sondes and acoustic equipment) can be used by the autosampler, before purchasing equipment.

When only basic parameters (such as stage) are needed to control an autosampler, it will usually be most cost effective to rely on built-in commercially available control systems that will not require extensive programming or the purchase of specialized accessories. Where multiple triggering and control variables are used in complex programming routines, it may be more advantageous to use a stand-alone control and data-logging system.

Stand-Alone Control and Data Loggers

Often, after a thorough examination of the goals of the study and characteristics of available equipment, a stand-alone control and data-logging system is needed to provide the flexibility, control, ease of access, and storage of data required for the project. These systems can record and utilize data from various independent sensors, such as water-quality sondes, to operate sampling equipment and can often provide remote access and programmability through cellular or satellite communications. A thorough cost-benefit analysis should be performed to determine if it is more cost effective to control the sampling equipment using a flexible stand-alone control system, or the (often more rigid) options available in a built-in control system. This analysis must not only consider cost, but also the time required to design and test specific computer code needed by stand-alone equipment. However, for many projects, once a “standard” control program is developed, it can easily be modified to the specifics of each sampled water source.

One benefit of stand-alone control systems is the ability to communicate and control the sampling equipment remotely via phone lines, radio, or cellphones; as often this allows the operating parameters of the autosampler to be adjusted or changed based on field conditions. In this regard, before proceeding with a stand-alone control system, the following should be considered:

After considering these questions, it may be advantageous to utilize stand-alone data loggers to measure river parameters, control the timing of sampling equipment, store sample and stream data, and report to remote locations the status of the sampling and river conditions. Numerous data-logging and control systems are commercially available that allow multiple stream and water-quality parameters to be stored and utilized for control of an autosampler, and to transmit information via radio or phone systems. The measured parameters, as well as data produced from a nearby USGS streamgage station, can be logged and used to trigger and operate the autosamplers. Data obtained from USGS equipment is available either as a numerical stream of digital data, transmitted via wire line or radio from the gaging equipment or, in some cases, as a direct-electrical signal from equipment such as pressure transducers (in a bubbling system, for example), stage or velocity (from radar or acoustic monitors), or parameters measured by water-quality sondes. However, it is imperative to notify and obtain permission from the relevant Science Center before undertaking a connection to existing USGS streamgaging station equipment.

Figure 15 shows a schematic of a system configured to use a stand-alone controller and data logger to operate two autosamplers; one autosampler collects discrete water samples needed to produce the time-sequence of conditions in the river, while the second autosampler produces a composite sample for chemical analysis. To produce a time- or discharge-weighted sample, the data logger monitors numerous stream parameters (such as stage), calculates and sums discharge, and starts sample collection at the desired intervals. Alternatively, the system could read several parameters from a water-quality data sonde (such as turbidity, specific conductance, DO, and temperature), monitor for the onset of precipitation, and begin sampling with the onset of pre-selected conditions. Any of such parameters can be used to initiate and control the autosamplers, and the data logger can be programmed with the relevant stage-discharge relation needed to track and initiate operation of the composite sample. In addition, the data logger records sample data (time and date) and equipment status, and if the system is equipped with a telemetry system (radio, satellite, land line, or cellular phone systems) allows the field crew to remotely monitor and if necessary, adjust or stop the sampling equipment.

Analyte-Specific Considerations

When selecting autosampler equipment, careful consideration should be given to the class of analyte(s) being studied. Many analytical parameters require that only certain materials contact the sample; materials made of substances that will not contaminate or alter the sample by absorption or leaching of chemicals. The components that contact the sampled water include the intake line, pump tubing, distributor line, tubing connectors, sample containers, and any check- or diversion-valve (if used). These components are typically made of metal, plastic, glass, or possibly ceramic. The amount of reaction between the material and the sample will depend on the material composition, the chemistry of the water sampled, and the elapsed time the sampled fluid is in contact with the material. In general, the softer or more flexible forms of plastic or metal are more reactive than the rigid forms, and the more polished a surface the less reactive the material tends to be (U.S. Geological Survey, 2014). Table 2-1 in chapter A2 of the USGS “National Field Manual for the Collection of Water-Quality Data” (NFM) is a good reference to use when selecting the proper material for a given set of analytes (U.S. Geological Survey, 2014, p. 16).

For more information on the preservation requirements and holding times for specific analytes, see the National Environmental Methods Index (NEMI) at http://www.nemi.gov or contact the laboratory that will be analyzing the samples. It is not within the scope of this report to list all analyte-specific considerations for all parameters analyzed by the USGS; however, some general groups can be discussed. Table 4 lists some of the major groups of constituents for which autosamplers are commonly used, some of the criteria for preservation, and some of the general holding times typical for each parameter group. For more specific information on each analyte of interest, contact the USGS National Water Quality Laboratory or the contract laboratory doing the analysis for the project.

Generally, for suspended sediment, major-ions, and nutrients, low-cost plastic lines and bottles can be used. For trace metals, only Teflon or Teflon-lined tubing and connectors should be used. It is often possible to replace the original metal parts in the autosampler, such as connectors, with plastic or Teflon parts. When sampling for trace organics, either Teflon (preferred) or stainless-steel tubing may be used. All parts of the autosampler, the intake lines, pump tubing and connectors that come in direct contact with a water sample should be cleaned by following USGS cleaning protocols before deployment (Appendix 2). If a permanent intake line is installed, then project-specific methods must be developed to ensure its cleanliness (or replacement) between sampling events.

Consideration must also be given to sample preservation and hold times, and the effort needed to shorten the interval between collection, preservation, and retrieval. Hydrologic event sampling is often conducted over days and weeks, and time-sensitive samples need to be removed and processed as quickly as possible after collection.

Special Sampling Situations

While most sampling conducted by the USGS involves natural systems (rivers, streams, lakes, estuaries, and groundwater), autosamplers are often used in special situations where sampling is conducted for non-typical analytes, such as pharmaceuticals or microbiological constituents. These studies may include sampling sewage from sewers, sewage treatment plant inlets and outfalls, and even septic systems. The composition of these wastes often varies greatly over the period of a day, and autosamplers afford a means to characterize the waste chemistries. Sampling raw and treated wastes requires special handling and safety considerations which are not discussed in this document, but with care, samples can be safely obtained using an autosampler. Help and advice in designing sampling systems can often be obtained from the environmental monitoring staff working at local public sewage authorities.

Additional Features to Consider When Designing an Autosampler System

The design of autosampler systems often include features that may prove useful or necessary to achieve the project goals. Some of these include

Selecting Location of Inlets in Rivers and Streams

In any study that uses an autosampler, it is the responsibility of the researcher to demonstrate that the intake point is located where a reasonably representative sample of the river channel can be collected. The velocity of water flowing in a stream varies within a channel’s cross section, with the highest velocity near the center of the channel (termed the thalweg). This results in a complex distribution of velocities across the channel, causing an inhomogeneous distribution of suspended sediment in the channel, both in the vertical and horizontal directions. Meanders and obstacles in the stream shift the thalweg position and cause turbulence in some areas, while other areas in the channel velocities may be much slower (even imperceptible). Input from tributaries and storm-water outfalls also affect the cross-channel distribution of dissolved components and sediment, and it is common to have inputted water travel a considerable distance downstream before the channel becomes well-mixed. For an autosampler to accurately characterize a stream, careful consideration must therefore be given to the placement of the intake.

The primary concern of the researcher is to determine how well a sample collected at a single point in the stream (the nozzle location) represents the water and suspended sediment in the entire cross section of the stream channel. A “best” or representative sample can be obtained by placing the autosampler intake at a point where the stream concentration approximates the mean cross-channel concentration of the parameters of interest, over the full range of expected streamflow. This goal has great merit, but to locate the point in the cross section where the mean cross-sectional concentration can be obtained (for each parameter of interest) may take considerable effort, and this point may change under different discharge regimes. Specific guidelines for locating the ideal intake position in a water body would have little transfer value among different reaches of a stream, or among different streams. For this reason, modifications have been applied to the findings of Edwards and Glysson (1999) and are presented as generalized guidelines in the following sections. Finding the optimal nozzle location in a cross-section is typically less a factor when sampling is for dissolved components, as turbulence in the flow usually results in good cross-sectional mixing for the dissolved phase. This may not be the case, however, for suspended sediment, where the cross-sectional distribution for sediment is controlled in large measure by the velocity distribution and velocity is seldom uniform in a channel.

Example of Determining Channel Cross-Sectional Homogeneity

An example of data collected to determine cross-sectional homogeneity are presented in table 5 and figure 16. Discharge, specific conductance, DO, and SSC were measured at 10-ft intervals across a 290 ft wide channel in the Passaic River, approximately ¼ mile downstream of its confluence with the Pompton River, located in northern New Jersey (U.S. Geological Survey, 2024). The measurements were made using a water-quality sonde and by collecting individual grab samples. Sampling was conducted during low-flow conditions with a water depth of approximately 3 ft. Preliminary cross-section measurements of turbidity showed the river was well mixed vertically, so multi-depth measurements and samples would not be required. A three-person field crew then began sampling: one person collected samples from EWI locations across the channel, the second made sonde measurements, and the third collected discrete grab samples immediately upstream of the intake point.

Table 5.    

Cross-sectional data from U.S. Geological Survey station no. 01389005 Passaic River below Pompton River at Two Bridges, New Jersey.

[Data are from U.S. Geological Survey (2024). Time given in military time. ft, feet; mg/L, milligrams per liter; °C, degrees Celsius; μS/cm, microsiemens per centimeter; ft³/s, cubic feet per second; Q, rate of discharge in flowing water; LEW, left edge of water; REW, right edge of water; %, percent; NA, not applicable]

Table 5.    Cross-sectional data from U.S. Geological Survey station no. 01389005 Passaic River below Pompton River at Two Bridges, New Jersey.

Discrete sample	Distance from left bank (ft)	Dissolved oxygen (mg/L)	Temperature (°C)	Specific conductance (μS/cm)	Suspended sediment (mg/L)
LEW at 1625	40	10.3	24.6	411	2
	50	10.6	24.6	412	2
	60	10.8	24.5	413	3
Point of left intake	70	10.7	24.3	416	9
	80	10.7	24.3	419	15
	90	10.7	24.3	431	16
	100	10.6	24.3	466	12
	110	10.6	24.3	508	17
	120	10.4	24.4	581	27
	130	10.4	24.5	668	43
	140	10.4	24.5	681	32
Point of middle intake	150	10.4	24.5	682	30
	160	10.6	24.5	718	22
	170	10.6	24.5	734	29
	180	10.6	24.6	734	36
	190	11.7	24.6	738	20
	200	10.8	24.6	737	19
	210	11	24.6	740	21
Point of right intake	220	11	24.7	737	17
	230	11.2	24.7	738	19
	240	11.3	24.7	737	15
	250	11.3	24.7	734	11
	260	11.4	24.7	734	13
	270	11.2	24.7	737	14
	280	10.9	24.7	739	9
REW at 1600	290	10.5	24.6	736	5
Sum	NA	NA	NA	NA	NA
Percent difference of mean from cross-sectional mean	1.9	0	11	−2.9	NA
Values in discrete cross-sectional grab samples
Maximum of section	11.7	24.7	740	43	9.98
Minimum of section	10.3	24.3	411	2	0
Percent difference1	12.7	1.6	57.2	182	200
Mean of section	10.8	24.5	630	18	5.67
Discharge-weighted mean values of section	10.8	24.5	671	20.4	NA
Median of section	10.7	24.6	726	16.5	6.09
Values in discrete samples taken from intake lines at 16:00
Right	NA	11.1	24.5	745	22
Middle	NA	10.3	24.5	630	28
Left	NA	10.5	24.4	427	12
Mean	NA	10.6	24.5	601	21

¹

Percent difference defined as 100% * (maximum – minimum) / [(maximum + minimum)/2]

Table 5.    

Cross-sectional data from U.S. Geological Survey station no. 01389005 Passaic River below Pompton River at Two Bridges, New Jersey.

[Data are from U.S. Geological Survey (2024). Time given in military time. ft, feet; mg/L, milligrams per liter; °C, degrees Celsius; μS/cm, microsiemens per centimeter; ft³/s, cubic feet per second; Q, rate of discharge in flowing water; LEW, left edge of water; REW, right edge of water; %, percent; NA, not applicable]

Table 5.    Cross-sectional data from U.S. Geological Survey station no. 01389005 Passaic River below Pompton River at Two Bridges, New Jersey.

Discharge (ft3/s)	Percent of total Q	Dissolved oxygen (%)	Temperature (%)	Specific conductance (%)	Suspended-sediment concentration (%)
0	0	0	0	0	0
0	1.39	0.01	0.1	0.23	3.89
1.39	0	0	0	0	0
0	7.96	0.05	0.58	1.31	22.5
7.96	5.98	0.04	0.43	0.99	17
5.98	5.25	0.04	0.38	0.87	15.4
5.25	4.68	0.03	0.34	0.77	14.8
4.68	6.19	0.04	0.45	1.02	21.3
6.19	6.79	0.05	0.48	1.12	26.8
6.79	9.86	0.07	0.7	1.64	44.7
9.86	3.62	0.02	0.26	0.6	16.7
3.62	8.46	0.06	0.6	1.41	39.2
8.46	9.05	0.06	0.65	1.5	44.1
9.05	4.52	0.03	0.33	0.75	22.5
4.52	7.4	0.05	0.53	1.24	36.9
7.4	9.98	0.07	0.79	1.67	50
9.98	8.69	0.06	0.64	1.45	43.5
8.69	6.88	0.05	0.51	1.15	34.6
6.88	5.88	0.04	0.44	0.99	29.4
5.88	7.8	0.05	0.59	1.31	39.1
7.8	6.23	0.04	0.48	1.04	31.2
6.23	5.52	0.04	0.42	0.93	27.5
5.52	4.16	0.03	0.32	0.7	20.7
4.16	8.67	0.06	0.66	1.45	43.4
8.67	2.37	0.02	0.18	0.4	11.9
2.37	0	0	0	0	0
147.33	1	10.8	24.5	671.3	20.4
NA	NA	NA	NA	NA	NA
Values in discrete cross-sectional grab samples
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA
Values in discrete samples taken from intake lines at 16:00
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA
NA	NA	NA	NA	NA	NA

¹

Percent difference defined as 100% * (maximum – minimum) / [(maximum + minimum)/2]

The cross-sectional data indicated that the channel was poorly mixed at this section. Velocity varied considerably among the section, with three areas where no velocity was apparent. The specific conductance data show a demarcation between low values (approximately 400 µS/cm) near the right bank and high values (approximately 700 µS/cm) at about 100 ft from the left bank; temperature and SSC were also high near the right bank. Dissolved oxygen concentrations (not shown in figure 16) were constant across the stream channel. This was not unexpected because of the shallow water depth.

On the basis of this study, it was deemed that three inlets would be required to adequately characterize the channel: one inlet located at 75 ft, another at 150 ft, and a third at 220 ft from the right bank. A system was installed that used a submersible pump equipped with an automatic valve that allowed each sampling line to sequentially deliver water to a plastic tub in the station. A continuous water-quality sonde was deployed in this tub and set to measure water parameters each time the tub was filled. When needed, an autosampler was set besides the tub, and intake line was installed to the tub. A stand-alone data logger was used to control the pump, valve, sonde, and autosampler.

The variability in the cross-section values can be quantified by calculating the percent difference between the maximum and minimum values for each parameter using the following formula for equation 1:

$\frac{\left(maximum-minimum\right)}{\left(maximum+minimum\right)/2}\text{*}100=Percentdifference$

(1)

The percent difference between the minimum and maximum values in the example cross section were found to be 1.6-percent for temperature, 57.2-percent for specific conductance, 182-percent for suspended-sediment concentration, and 200-percent for discharge. This demonstrates that the cross section is not well mixed for dissolved ions (specific conductance), suspended sediment, and discharge during periods of low flow. Nearly identical results were obtained from measurements made later at higher discharge conditions.

Subsequent sampling was conducted to determine if the three intake lines were placed in appropriate locations in the cross section to produce an accurate “average” sample. Water collected from the right, middle, and left intakes were compared to the flow-weighted cross-sectional mean values. Cross-sectional flow-weighted mean values were calculated to be 10.8 milligrams per liter (mg/L), 24.5 degrees Celsius (^oC), 671µS/cm, and 20.4 mg/L, for DO, temperature, specific conductance, and suspended-sediment concentration, respectively. Mean values in the samples collecting through the three intake lines were 10.6 mg/L, 24.5 ^oC, 601µS/cm, and 21.0 mg/L for DO, temperature, specific conductance, and suspended-sediment concentration, respectively. These values correspond to percent differences of 1.9-percent for DO, 0 percent for temperature, 11-percent for specific conductance, and −2.9-percent for suspended-sediment concentration. Of the three inlets, the right inlet most closely approximates the mean DO and suspended-sediment concentration, while the middle most closely resembles the specific conductance. These results demonstrate that while cross-sectional variability is high, the intake lines are positioned such that the average sample closely resembles the flow-weighted mean of the cross-section.

Intake-Line Construction

After the intake location has been chosen, consideration must be given to its design and construction. Because of the myriad of configurations possible, only general suggestions are provided here.

In its simplest form, the intake line is a plastic or metal tube that extends out into the channel from the bank side or sits near the stream bottom and extends up into the flow, with its nozzle orientated in the down-flow direction. More complex installations can be devised that incorporate not only inlet lines but water-quality sondes (fig. 17). The intake line should be kept as short as possible; the shorter the line, the less power will be used to draw in a sample and to flush the line, thus reducing power usage from batteries and wear on the pump tubing. A short inlet line also reduces pumping head-loss and thus helps maintain water velocity in the line, thereby minimizing potential for sediment to deposit in the line. If left in the river, then the exposed portions of the tube line may become stained with oxides, oil, and algae (fig. 18) and will need to be either cut off and extended, or the entire line replaced before each sampling event.

Between the station shelter and the stream, the intake line will need to be secured against flood water, the impact of floating debris, and from vandalism (fig. 19). Burying the intake lines (encased in an armoring conduit) protects the lines from high water and places them out of sight from vandals. Once in the stream, the lines will need to be protected from trees and other objects moving through the water during floods; this can often be done using gravel or boulders taken from the stream, or riprap brought to the site. However, the intake armoring line should be designed to have minimal effect on the natural flow of the stream. As much of the armoring conduit should be buried in the streambed as possible. Steel pipes are good armoring conduits and supply more weight and stability than plastic pipes. If using steel pipes, ensure that any burs in the pipe are removed.

A typical installation places the intake line at a single fixed point in the cross section, generally near the bottom in the deepest portion of the stream towards the area of highest cross-sectional velocity. This allows the intake to remain submerged during all streamflow conditions. If the intake needs to extend far into the stream, the line should be run along the stream bottom with the armoring pipe diameter being kept as small as possible. The lines can be held in place with rebar stakes or short steel fence poles driven into the streambed, and then covered with large riprap or sandbags. Care should be taken that the inlet line does not kink where it enters a structure, extra tubing length should be provided to keep bends smooth.

The intake nozzle should be raised up sufficiently so that it remains above the bed-sediment during high streamflow, thereby collecting only suspended materials and not coarser bed materials moving along the channel bottom. As much as possible, the lines should slope continually down-hill from the autosampler to the nozzle, without “dips” or low spots where sediment will accumulate. Bends where water sediment can accumulate should also be kept to a minimum, but if unavoidable, should be limited only to the ends of steep sections (down the riverbank). Where the intake line changes direction in the stream, for example, at the stream’s edge and at the nozzle, bends should be gradual, using sequential 45-degree bends. Angling the line diagonally out into the stream will help minimize the final bend used to direct the nozzle downstream. It is advantageous to install multiple intake lines, each set at different depths in the stream, to facilitate sampling over all stages (low to high). In using multiple inlet lines, the selection of the line to be used can be manual, or each line can have a dedicated autosampler, or a data logger and control system utilizing automatic valves can be deployed. Redundant lines also help to ensure that a storm event is successfully sampled if one line fails.

Armoring

In permanent or semi-permanent deployments, the intake line is typically armored (fig. 20) to protect it from damage during storm flow. Inlet lines can be armored using a steel pipe, or plastic tube (PVC; flexible or stiff) such as continuous 1–2 in. plastic waterline. Armoring should run from the sampling structure to the stream edge, then to near the inlet nozzle location. Using a continuous length of black plastic water line, with few or no connectors, allows the intake line to be easily removed and replaced (fig. 21). Between the sampling shelter and the stream bank, the line can be buried or covered with stone riprap. Within the channel, the lines can run on or be buried under the stream bottom, and kept in place using stream boulders or gravel, riprap, steel stakes (rebar or fence posts), or concrete anchor blocks. Used steel pipe (2–3 in. diameter) is an economical armoring tube and works well for armoring the lines in high velocity streams, because their extra weight helps keep them secured in the streambed.

Intake Orientation

The orientation of the intake nozzle will affect the sampling efficiency of an autosampler. There are five directions in which an intake can be oriented to the streamflow, as shown in figure 22: A, perpendicular to the stream bank, then turning and pointing upstream, B, perpendicular to the stream bank and to streamflow, C, perpendicular to the stream bank, then extending vertically upward into the water, D, perpendicular to the stream bank then extending vertically downward into the water, and, E, perpendicular to the stream bank, then turning and pointing downstream. Of these five, the orientations shown in figures 22A, 22C, and 22D should be avoided because they are known to cause high sampling errors and have problems with debris collecting on the intake (Edwards and Glysson, 1999). The orientation shown in figure 22B, with the intake positioned perpendicular into the streamflow, is the simplest to construct. However, with this orientation, sand-size particles may not be adequately captured (Edwards and Glysson, 1999). It should be emphasized that sharp turns in the inlet lines should be avoided as these often become clogged with sediment; all turns should be kept gradual and smooth.

Winterstein and Stefan (1983) and Winterstein (1986) found that sampling efficiency was highest when the intake was pointed directly downstream (fig. 22E) and decreased as the nozzle orientation was shifted towards a horizontal configuration (fig. 22B). Winterstein demonstrated that when pointed downstream, the intake causes a small eddy to form that envelops suspended particles, helping to obtain a representative sample of the suspended load, especially the coarser grain-sized material. A sampling efficiency of about 0.90 was achieved for all sands tested (Winterstein, 1986).

Winterstein (1986) further demonstrated the importance of the downstream configuration by testing the sampling efficiencies of intakes oriented at 45 degrees either side of pointing directly downstream. When positioned diagonally with respect to flow at 135 and 225 degrees, (with 180 degrees being directly downstream), sampling efficiency varied from 0.39 for 0.20 mm sized sand to 0.84 for 0.06 mm sized sand using an isokinetic withdrawal rate, and from 0.70 for the 0.20 mm sand to 0.90 for 0.06 mm sand at a rate that was twice the isokinetic rate. With an inlet line oriented diagonally across flow, a higher pump rate was necessary to match the sampling efficiency of a downstream orientation. The FISP studied the effects of non-isokinetic sampling with a nozzle (intake) pointed directly into the streamflow (Tennessee Valley Authority and others, 1941). The results of that study showed that the errors were negligible for sizes less than 0.06 mm.

Based on this information, Winterstein (1986) concluded that the size distribution of suspended-sediment samples collected by an autosampler with the intake set at any angle other than 180 degrees (directly downstream) to the streamflow will be biased towards the smaller sediment sizes. Particles in the silt-clay range will be only slightly under-represented while the sand-size material may be greatly under sampled. The bias increases with increasing particle size. It should be noted that none of the tests run by Winterstein (1986) used multiple-hole strainers placed over the intake. The use of a strainer may further affect the efficiency of the autosampler and its ability to collect a representative sample. If a screen is used, then it is instructive to compare samples collected with and without the screen to document the effect of the strainer. For more information on the orientation of the sample line, see Edwards and Glysson (1999)

Intake Nozzle Design

The intake point, or nozzle of the line, can simply be the tube end set at a fixed height above the channel bed. However, more elaborate designs can be used to maintain the intake nozzle at the same relative depth below the water surface. Similarly, the nozzle end can be left open or have a screen or strainer attached to the line.

Eads and Thomas (1983) demonstrated that an intake nozzle set at a fixed level will collect material from different portions of the flow profile, which changes as the stage and flow velocity change. They developed a depth-proportional intake device that maintains the intake nozzle at the same proportion of the stream depth regardless of stage (fig. 23). The device consists of a boom, an anchor on the streambed, a pivot, and a float. The unit is fixed in the streambed using an anchor, with the boom and float facing downstream. This design helps wash debris off the boom by the force of the flowing water. This device was originally designed for relatively small streams or culverts, as the boom length needs to be at least twice the maximum expected depth of the stream. The intake is attached to the boom at a point 0.60 times the distance from the pivot to the float; this allows the intake nozzle to remain at 0.60 times the water depth. This device is deployed with the orientation shown in figure 22E.

For streams carrying a high sediment load (especially with sand), it is sometimes advantageous to attach a static mixer at the end of the intake line, as shown in figure 24. The mixer helps to (1) provide a secure mount for the intake line, (2) reduces the velocity of the water around the tube end, thereby aiding the pump suction to capture particles, and (3) helps mix the sediment in the water being sampled (Smith, 2002). Static mixers are especially helpful when sampling culverts, pipes, concrete channels, and ditches. Smith described this static mixer² as follows:

“The static mixers were constructed from a 0.5-in. marine-grade homogenous polymer sheet and consisted of two semicircular plates 1.2 in. high at the center. Two plastic 0.5-in. bulkhead-compression fittings were attached to each side of the first plate 1.2 in. from the center and 0.6 in. from the bottom. The second plate had two semi-circular 0.7-in. holes in parallel with the bulkhead fitting to prevent sediment accumulation between the two plates and was mounted about 4 in. behind the primary plate.” (Smith, 2002, p. 14)

Line Removal and Replacement

It is often necessary to remove the intake line for cleaning, replacement, or to collect quality-assurance samples. The ability to easily replace the intake line is one of the paramount considerations when designing an autosampler system; many different designs and tricks have been used to simplify the task. For example, the USGS Wyoming-Montana Water Science Center developed a line-and-pulley system to help in removing the intake and tubing without the need for the field crew to enter the water (fig. 25). It consists of a plastic armoring pipe with a smaller tube attached along the outside. The intake tubing is fed through the armoring pipe using a rope attached to the end of the intake line; this rope extends back to the sampling station. A second rope is fed from the nozzle back to the station through the exterior tube. Pulling on the rope fed through the armoring brings the inlet line back into the station, while pulling on the line fed through the outer tube returns the nozzle end back to its original position in the stream.

Special Considerations—Locating Inlets in Culverts and Pipes

Installing inlet lines in culverts and storm drains, a common task for using an autosampler, can be difficult due to the extreme flow velocity that can occur in the pipe; it is typical to have flows that rapidly cycle from zero to maximum velocity in minutes. Typically, a single intake line is used with the nozzle placed near the bottom of the pipe, with the intake line anchored to the pipe. The line is typically dropped vertically inside a box or manhole and may be anchored to the walls of the structure and to the top of the pipe. Alternatives are to attach the intake line to a heavily weighted cable hung into the pipe, or purchase or construct a metal “spider” to hold the inlet line within the pipe. In all deployments in pipes, the intake nozzle should be pointed down flow. Any wire, supports, or weights should be epoxy coated, as concrete or uncoated metal weights will react with the liquid and potentially degrade and become lost, or alter the chemistry of the liquid. In all buried pipes, extreme care must be taken to ensure that any weights or fixtures placed in the channel cannot become wedged in the pipe by debris.

Because water velocity can be extremely high in storm-water pipes, considerable effort may be required to secure the intake line in the pipe. In larger diameter storm-water pipes, the inlet lines are likely to be lost and will need to be replaced multiple times over the course of a study. Also, sediment loads may be extremely high during discharge events. Because coarse grain-sized sediment such as road grit will migrate along the floor of the pipe, sample bias will occur if the intake is not located near the middle of the pipe. In these situations, consideration should be given to a static mixer or a constant-depth fixture to keep the nozzle above the bottom of the storm drain (Barbaro, 2001).

Othmer and Berger (2002) devised a fixture that keeps the inlet nozzle at a set distance below the water surface but off the pipe floor in a partially filled pipe (fig. 26). This system floats the intake at constant depth below the liquid surface.

Types of Quality-Control Samples

A list of QA/QC sample types used to verify representativeness are shown in table 8. QA samples and procedures for bacteriological samples are listed in table 9.

“Laboratory” or “field equipment blanks” are used to document bias caused by poor equipment cleanliness or crossover contamination between samples. These blanks are collected by processing water that is free of the analyte(s) of interest using the same procedures used to collect environmental samples. These blanks can be produced in the laboratory before leaving for the field or can be prepared in the field. Generally, a “field equipment blank” is preferred, as it processes blank water through all steps of collecting and processing samples in the ambient field environment. For an autosampler, this means passing certified blank water through intake lines, pump and distributor, sample-collection bottles, and any compositing, splitting, and filtration procedures conducted (in the field or laboratory). If needed, these steps can be broken down to assess contamination caused by individual pieces of equipment or isolated steps in the collection process. Collecting equipment blanks for intake lines requires additional effort and must be well planned (see “Preparing Blanks of Intake Lines”). Prior to producing a laboratory- or field- equipment blanks, all parts of the sampling equipment should be cleaned and handled in a manner identical to that used for when preparing to collect field samples. Field blanks are processed and preserved along with field samples.

“Spiked samples” are defined as samples (either native water or certified clean water) that are fortified with a known concentration of a specific analyte or analytes. The observed concentrations of analytes in the spiked sample are compared to the expected concentrations, calculated from the spike-analyte concentration in the environmental sample (if used, or the certified concentration if blank water is used), and the mass of the analyte in the spiking solution added. Spikes are used to estimate bias in analytical methods, either from matrix effects caused by substances in the natural water, and any degradation of analytes that occur during the sample process and storage. Spiking solutions with certified concentrations can be obtained from an analytic laboratory or purchased from commercial chemical suppliers. When preparing spiked QA samples, three samples are needed: a non-spiked environmental sample, a replicate non-spiked environmental sample, and the spiked environmental sample. Because sample containers are often left open in an autosampler prior to, and for a short period following sample collection, the use of spiked samples is strongly recommended when using an autosampler for semi-volatile compounds (organic compounds, mercury, and lead), biologically active compounds, and bacteria.

“Replicate samples” are collected to document variability caused by collection methods and by analytical methods (analytic variability). Replicates can be collected in multiple ways using an autosampler. “Split replicates” can be collected from an autosampler by splitting a single or composite sample into replicates. A common technique to collect replicates by splitting one container of water using a churn or sample splitter device. “Sequential replicates” can easily be collected using an autosampler by programming it to collect consecutive samples at each sample initiation signal.

“Concurrent replicates” are multiple samples collected at the same time and place using different sampling techniques. For autosampler deployments, concurrent replicates play an important role, as they are used to compare samples collected by the autosampler and traditional cross-sectional isokinetic methods (EWI or EDI cross sections). Typically, the autosampler is programmed to collect discrete samples at appropriate time intervals (for example, every 10–15 minutes) over the period of time grab samples are being collected for the composite. The samples collected by the autosampler and manual methods, can be analyzed individually, or composited before the comparison is made. The “concurrent replicate” samples can be collected at a low flow condition, or during storm flow; however, conditions in the stream may change rapidly during a storm, thereby complicating the comparison of individual samples. “Concurrent replicate” samples are used to generate box-coefficients that allow corrections to be made between the results obtained by the two sampling methods. An example of box-coefficient calculations is provided in appendix 3.

Preparing Blanks of Intake Lines

Collection of a field equipment blank must include consideration of the intake line, thereby requiring the nozzle end to be accessed and placed into a container of certified blank water and pumped up to a collection bottle in the autosampler. It is important that the lines being blanked are first cleaned, either in the laboratory or the field. If this is not possible, then at a minimum the lines must be first flushed with clean water before preparing the blank. Cleaning is described in “Intake-Line Maintenance and Cleaning.” Flushing should use a minimum of three-tubing volumes of tap water followed by three-tubing volumes of distilled water (DSW), prior to introduction of certified blank water. As a general rule, equipment blanks should be prepared on the intake lines prior to each sampling event, or whenever a sample line is replaced. It should be noted; however, that intake line cleaning and blanking is constituent dependent, that is, for some constituents such as suspended sediment, a lower effort of cleaning or blanking is suitable. However, other constituents, such as trace metals and organics, will require a greater frequency and effort.

Holding-Time Studies

Holding-time studies may be necessary to determine if there is a constituent concentration change over the time samples are held in an autosampler before preservation steps are taken. This can be evaluated during a site visit by collecting a sample (with the autosampler) and immediately splitting it into two sub-samples. One sub-sample is preserved, processed, and shipped to the laboratory for analysis, while the other sub-sample is left in the autosampler until the next scheduled site visit when it is retrieved, processed, and sent to the laboratory.

This type of study is especially necessary when samples will be exposed to elevated summer temperatures in a non-climate-controlled sampling station. It may be necessary to deploy a temperature logger in the autosampler to track changes the samples are being exposed to, even if the samples are being refrigerated in the autosampler. Any exposure of open sample bottles to high temperatures can speed up the degradation and volatilization of dissolved constituents and can increase the growth rate of bacteria.

Quality Assurance Samples for Trace Components

QA/QC samples (field blanks and field equipment blanks) are especially important when using autosamplers to collect samples for trace inorganic or organic compound analysis. Unlike grab samples, when an autosampler is used, the sampling lines (inlet, pump, and distribution tubes) may not be “single use” and therefore, may not be considered “clean” after the equipment is deployed and the first sample is collected. Field equipment blanks take on special significance because they document that the equipment was properly cleaned and not compromised during deployment. Likewise, field blanks and spiked blanks are essential for documenting the effects of cross-bottle contamination and air-borne contamination on trace component concentrations. Therefore, it is advisable that when sampling for trace components the number of QA/QC blanks (field and equipment blanks) is increased to 25-percent or more of the total samples, distributed equally across all sampling deployments. The following steps are useful to help ensure successful trace component sampling:

Quality Assurance for Bacterial Constituents

Autosamplers are often used to sample rivers and waste effluents for bacteriological constituents. Generally, they should be employed only where very high bacterial values are expected, as cross-contamination is a significant concern; it is very difficult to clean and sterilize all parts of the equipment sufficiently to allow for repeatable low-level analyses. Adding extra purge-rinse cycles or using a continuously pumping system including a bypass valve on the sampling line may help ensure that each sample is collected from a well-flushed and purged intake line. However, continuous flow of water may still allow some bacterial growth to occur in the lines, especially in streams where algae is abundant. Chilling of the samples while in the autosampler is also very important, either by ice or by using a refrigerated autosampler. Aulenbach (2009) has conducted studies on the holding times of bacteria for up to 62 hours for fecal coliform by mFC agar method, and total coliform and Escherichia coli; however, it was found that samples held for 18 hours provided results similar to those found in samples analyzed after the recommended 6-hour holding time. However, they stressed the importance of maintaining proper sterile condition during collection, preservation, storage, and analysis of bacteriological samples.

The following QC procedures are suggested for sampling for bacteriological analysis:

Types of Sampling Schema

The principal types of sampling schema are based on whether samples are to be collected at a fixed time or discharge interval and produce either discrete or composited samples.

Scheme for Collecting Discrete Samples

Autosamplers are extremely beneficial when multiple, discrete samples need to be collected at a preset increase in stage or a preset discharge interval. Examples include the studies where a sample is needed when flow first begins in a culvert, storm-water discharges from a pipe, or maximum stage is reached in a river. Automatic equipment removes the need for humans to evaluate the stream conditions and initiate (or terminate) the sample collection. A “discrete sample” may consist of a single or multiple bottles filled sequentially for each triggering signal sent from the controller. Multiple-bottle discrete samples are used when the analytes or classes of analyte cannot be measured from a single bottle of water. A common example is sampling for dissolved and total analytes, or when samples are needed for suspended sediment and particulate organic matter analysis. Collecting individual, discrete samples maximizes the amount of sediment or carbon in each bottle available for analysis and removes the chance for bias arising from incomplete homogenization when a single sample is used for analyses of multiple constituents.

Discrete samples (or aliquots) collected at preset time intervals provide information on the variation in stream chemistry. Figure 27 shows an example of the timing of sample collection when an autosampler was programed to collect discrete samples at two-hour intervals from a river during a hypothetical storm event that lasted three days. Note that because of the rapid increase in stage, only three samples were collected during the rising limb of the hydrograph, and 21 samples were collected later during the receding limb. If a shorter time interval was selected, more samples would have been collected earlier in the storm, but because autosamplers hold only a finite number of bottles (typically 24), less of the receding limb would have been characterized (unless field crews were available to replace bottles). Complete river condition data for this hypothetical storm are provided in appendix 4.

Discrete samples or aliquots (to be composited) can also be collected on fixed intervals of discharge passing the sampling station. Figure 28 shows the timing when samples would be collected during the same storm event using a preset discharge interval of 10 million gallons. In this scenario, more samples are collected during the rising limb and crest of the storm, than would be collected using a fixed-time interval approach. However, if too small a discharge interval is chosen, samples may overlap during the peak and therefore will be missed. Also, while the sampling will better characterize the rising limb and crest of the storm hydrograph, fewer samples will be collected during the receding limb and important information characterizing water entering the river from tributaries far upstream in the watershed may be missed. It should be reemphasized that only the flow-weighted sampling scheme produces an unbiased “average concentration” for the storm.

Event-Based Sampling

The main advantage in deploying an autosampler is the ability for it to monitor and automatically respond to changing flow conditions, and then to begin sampling without direct human oversight. This type of sampling is termed “event-based sampling”; after the initiation of sampling (the occurrence of the event), sampling may be undertaken either on a discrete time- or flow-basis. In this document an “event” refers to a prescribed change in environmental conditions. Typically, an autosampler is set to continue sampling in a preset manner only while the environmental conditions remain within set boundaries.

The most common example of event-based sampling occurs when a stream is sampled in response to the onset of precipitation. Because autosamplers hold only a finite number of bottles (or a fixed-bottle volume), sampling may stop before the “event” has concluded. In contrast, time-based sampling allows the researcher to be aware of the timing, the number of samples collected, and when the sampling will be concluded. For example, in a monitoring program that collects samples weekly from a river, the researcher will know precisely when the autosampler will need servicing and the sample bottles replaced. As discussed below, “event-based” sampling requires much pre-planning and other considerations; these factors impact the design, servicing, cleaning, and other QA/QC aspects of the autosampler deployment.

Typical Autosampler Cycle for Collecting a Sample or Aliquot

When an autosampler receives a triggering command, the sample line is first purged with water. This purge is accomplished by reversing the direction of the peristaltic pump. The inlet line is next rinsed with stream water, re-purged, a cycle that can be repeated multiple times. The sample is then pumped into the sampler container. The line is then re-purged and the distributor system is moved to the next sample container and readied for the next trigger signal when this sampling cycle is repeated. This purge and rinse cycling can easily be performed if the autosampler is equipped with a liquid-detector that senses when the liquid has reached the pump head. It is recommended that autosampler programs include at least one rinse and one purge cycle before each sample.

Alternately, if a fluid-detection monitor is not available, the sampling system can be equipped with a valve system located past the peristaltic pump head but before the sample container (fig. 29). This allows the line to be flushed with native water without the need to reverse the pump. The valve is opened by a command from the data logger as the pump is started and allows water to be pumped to waste for a set time period. The valve is then closed, and the water is allowed to flow to the sample container.

Although more complicated, there are two advantages to the continuous flow system over the purge-and-rinse system:

• • With the purge-and-rinse system, rinse water stops in front of the pump head at the liquid detector and does not rinse the pump tubing.

• • With the purge-and-rinse system, debris and native water may not be completely expelled from the inlet tubing during the purge cycle. With the valve system the continuous flow of water helps keep the line clear of settled debris.

The valve system if often employed when an in-stream submersible pump system is used to bring water up to the autosampler. The submersible pump can be run continuously or at time intervals, and a valve is operated using a relay operated from the output signal of either a data logger or through the control port of the autosampler. The manufacturer of the autosampler or data logger should be consulted for the specific connections needed to operate a valve system.

Scheme for Collecting Composited Samples

Composited samples are collected to produce a sample that represents the “average” composition during a specific hydrologic event or a set time interval. Figure 30 shows the makeup of different types of composited samples. Two of the more commonly used composite samples are flow-weighted and time-averaged. These samples are produced by collecting an aliquot of stream water each time a triggering signal is received, then combining it with the previously collected aliquots. A flow-weighted composite sampling scheme is used to determine an average concentration of a constituent (such as sediment) moved during a storm or flood event. This average concentration can then be multiplied by the number of storm events, or the yearly length of time storm flow occurs, to obtain the yearly constituent load. Time-averaged composite samples are commonly collected for regulatory monitoring of discharge, for example, when monitoring to determine the daily outfall loading of nutrients from a sewage treatment plant. Regardless of the composite type, each aliquot should be collected at the specified interval and contain identical sample volumes (allowing for a reasonable margin of error).

Multiple-bottle composite samples can be collected by having the autosampler place identical volume aliquots sequentially into two or more bottles upon each triggering signal. Flow-weighted compositing provides a representative sample only if an accurate stage-discharge relation is available for the station, and only if an appropriate discharge interval has been chosen.

Multiple-bottle composite-samples are collected when multiple classes of parameters are to be measured that cannot be measured in a single sample; for example, when concentrations of inorganic or organic constituents are required along with suspended-sediment concentrations. As in discrete sampling, suspended sediment is best measured in its own sample.

To properly produce a flow-weighted composited sample, a system to measure, calculate, sum, and record stream discharge is needed. Stage data can be obtained at a nearby existing streamgage, or stage can be measured using a hydrostatic pressure transducer placed in the stream bottom and connected to the autosampler or data logger. In the latter case, a stage-discharge relation will need to be developed for the site. In general, flow-weighted composite samples can only be produced where an accurate stage-discharge relation has first been established. The autosampler or data logger evaluates stage, calculates discharge, sums the discharge, and initiates sampling when the appropriate volume is reached; the data logger also measures the volume of each aliquot collected using pump time, peristaltic pump rotations, the use of a flow sensor, or measured weight of the aliquot. Less sophisticated equipment is needed to produce a time-averaged composite sample; typically, only a timer is used to control the pump operation.

Collecting Large-Volume Composited Samples for Suspended-Sediment Phase Analysis

Autosamplers are particularly useful for obtaining large-volume samples from which sediment is removed for analysis of nutrients and trace organic contaminants. This approach, which is typically performed during storms, is common in total maximum daily load (TMDL) studies to provide data for hydro-geochemical modeling and is used to model the effects of reduction of nutrients (such as phosphorous), particulate carbon, and hydrophobic trace constituents such as pesticides. This type of modeling requires discharge and “realistic” sediment and contaminant concentrations for tributaries and outfalls entering the main-stream or rivers, lakes, and other water bodies. Only accurately produced input data, derived from accurate sampling, can produce realistic modeling output.

Whereas dissolved constituents can be detected at extremely low (sub-nanograms per liter) concentrations in small volumes (typically 1 liter or less), obtaining data at useable detection levels of sediment-bound trace constituents requires a large mass (10 grams or more) of suspended sediment. This is because the detection levels for sediment-bound concentrations are a function of the mass of sediment used in the chemical extraction process. In contrast to bed sediment, which can easily be obtained in hundreds of grams, the acquisition of sufficient mass of suspended sediment requires the collection of very large volumes of water, especially when suspended-sediment concentrations are low (less than 10 mg/L) during lower flow regimes. The sediment in collected water is typically filtered in the laboratory, and the sediment loaded filters sent for analysis.

Planning is needed to maximize the chance that sufficient sediment is collected during a storm event. It is of course, impossible to predict the discharge levels for an upcoming storm event, but by using historic data for the stream of interest (where available) or data from a nearby stream of similar size, the chance of obtaining sufficient sediment is increased.

Several issues need to be considered when designing a large-volume sampling program. First, it will involve collecting, transporting, and processing very large volumes of water (often up to 100 L); either requiring multiple trips to the sampling station to swap-out collection vessels of manageable size, or transporting large volumes of water from the field. Secondly, the sampling scheme will need to balance two opposing factors: sampling over as much of the storm hydrograph as possible while focusing on obtaining water containing high amounts of suspended sediment. Early in a storm event, it is common to find very high concentrations of suspended sediment; this is the sediment that has accumulated in the streambed since the last storm event. Sampling late in the storm will incorporate sediment derived from tributaries located far up in the watershed. Water late in a storm will typically contain low concentrations of sediment, so collecting this water will add very little sediment mass to the sample. A balance needs to be struck between the coverage of sampling during a storm, and the mass of sediment collected. Complicating this is that in flashy urban streams, storm flows often last only for a few hours.

Keeping these factors in mind, sampling schemes can be devised that will help ensure sufficient sediment is obtained so that representative concentrations of trace analytes can be determined in suspended sediment and POC; data needed to provide accurate load information needed for TMDL modeling. Two examples help illustrate the relation between sampling scheme and mass of sediment ultimately obtained; these examples use data collected during sampling conducted on the Northeast Branch of the Anacostia River (fig. 31) and Watts Branch, (fig. 32) both located in Washington D.C. A pair of autosamplers were used in this study, one to collect paired discrete samples for suspended-sediment concentration and POC (throughout storm events), and a second to produce a composited sample for sediment analysis. The discrete samples were collected each time an aliquot was obtained for the composite sample. The target volume for the composite sample was 75 L and was collected in a Teflon drum liner held in a covered plastic tub (fig. 9). A water-quality sonde was deployed near the inlet nozzle to measure turbidity and specific conductance. For the Northeast Branch, a relation between turbidity (T), discharge (Q), and suspended-sediment concentration (SSC) developed by Miller and others (2007) was used to predict suspended-sediment concentration using equation 5:

Log

₁₀

SSC = log

₁₀

T × 0.9621 + log

₁₀

Q × 0.1710 − 0.2531

(5)

A similar model was developed for Watts Branch. This type of model is often not available; however, even an approximate relation based on a few historic suspended-sediment concentration and discharge values can aid in developing a sampling scheme.

Using a 0.5 ft rise in stage as a trigger, different sampling schema were evaluated to determine the mass of sediment that would be obtained in a large-volume composite sample (75 L, maximum) along with the coverage of the storm hydrograph.

For example, choosing a 10 million-gallon discharge interval and collecting 1 L aliquots, the sampling interval and stream parameters are depicted in figure 31. This interval resulted in collection of 33 aliquots; the suspended-sediment concentration in each aliquot collected and estimated using measured turbidity and discharge, would result in obtaining approximately 2.4 grams of sediment, sufficient for chemical analysis requiring a minimum of 1 gram of sediment. Importantly, over 95-percent of the total storm discharge (32.33 million gallons [Mgal] of the 34.07 Mgal passing the station), and 99-percent of the total sediment load (90.0 grams of the 90.9 grams passing the station) would be represented by the aliquots. This example illustrates how an existing hydrograph, combined with a suspended-sediment-turbidity model, can be used to determine a starting point for a sampling scheme that will provide suitable mass of sediment for chemical analysis and coverage of a storm.

Predicting a suitable sample interval for smaller, flashy streams where storm flow typically lasts less than a day, is more difficult. Here, successful sampling (obtaining a suitable mass of sediment that is representative of the entire storm hydrograph) becomes constrained by the time required to pump a volume of aliquot (which is a function of the pumping rate) and the duration and the number of purge and rinse cycles. An example is presented in figure 32, which shows the hydrograph for a storm during December 5–7, 2016, on Watts Branch, a flashy urban stream draining an area in southeastern Washington D.C. (Wilson 2019). The suspended-sediment concentration and resulting mass of sediment in the stream were calculated using the turbidity, discharge, and suspended-sediment concentration model from Wilson (2019). The estimated mass of sediment that would be obtained using several different sampling schemes are presented in table 10; these schemes include sampling every 10 minutes (time composite sampling) or every 0.5 million liters (ML) of discharge (flow-weighted composite sampling) and collecting either 1 L or 2.5 L aliquots for different number of bottles. Using these calculations, the mass of sediment, and the percentage of the total discharge and total sediment load that would be represented by the sampling were calculated. Flashy streams are difficult to sample because if a small discharge interval (such as 0.5 ML) is chosen, the sampling may occur at extremely short time intervals (in this case, as short as 2 minutes); this short of an interval is unlikely to be possible given that several minutes will be needed to rinse and flush the sampling line. In other words, choosing a sampling interval of 0.5 ML of discharge would result in samples being missed during the times of maximum flow (in this case, anytime discharge was over 95 cubic feet per second). This example exemplifies one advantage to keeping the inlet lines as short as possible, reducing the time needed for the purge/rinse/sample cycle between samples helps ensure that samples are not missed during periods of extreme discharge.

Table 10.    

Summary of sediment mass collected during the storm hydrograph shown in figure 32, using different trigger points, sampling intervals, and aliquot volumes.

[Data are from Wilson, 2019. ft, foot: NTU, nephelometric turbidity unit; ML, million liters; L, liter; g, grams]

Table 10.    Summary of sediment mass collected during the storm hydrograph shown in figure 32, using different trigger points, sampling intervals, and aliquot volumes.

Triggering point1	Sampling interval2	Aliquot volume, (L)	Number of aliquots obtained	Total sample volume obtained (L)	Percent of total flow covered by the sampling	Percent of total sediment mass covered by the sampling	Estimated sediment mass obtained (g)3
0.5 ft rise	10 minutes	1	60	60	94.2	98.7	6.1
0.5 ft rise	0.5 ML	1	48	48	75.2	86.6	7.6
100 NTU	0.5 ML	1	48	48	76.8	87.9	7.6
100 NTU	0.5 ML	2.5	24	60	36.8	32.7	2.9

¹

Triggering point refers to change in river conditions that initiate sampling. An aliquot is taken at initiation.

²

Sampling intervals are the variables determining when each aliquot is collected. Intervals of river discharge are in millions of liters. Sampling at a constant discharge interval produces a flow-weighted sample, while sampling on set time intervals does not. The shortest interval possible with the equipment being used was four minutes, and at this short time interval, the autosampler misses aliquots during times of peak discharge.

³

The mass of sediment collected was calculated using an equation relating historic turbidity measurements and suspended-sediment concentration in discrete grab samples collected between 2000 and 2016.

As shown in table 10, however, four of these sampling schemes would cover more than 75-percent of the storm discharge and 87-percent of the mass of sediment that was ultimately transported during the storm event and would provide over 6 grams of sediment for chemical analysis. By increasing the volume of each aliquot, thereby decreasing the number of total aliquots required by the volume of sample container held by the autosampler, a suitable mass of sediment would be obtained, but the sampling would cover only about 33-percent of the total water volume and sediment mass transported during the storm. This also illustrates the tradeoffs that need to be considered when designing a large-volume sampling program.

Composite Sampling of Sewers and other Waste Streams

When using an autosampler to produce composite samples from sewers and other waste streams, particular attention must be given to the design of the sampling program. The flow in these waste streams is highly variable and can occur as many “pulses” of water originating from different upstream discharges to the sewer pipe, some very close to the sampling point and some far “upstream” from the sampling point. At the inflow to a large sewage treatment plant, the pulses inherent in a waste stream become mixed, but even still, the flow and composition can change over the course of a day (high discharge in mid-morning and early evening, low discharge in between). The flow, therefore, cannot be assumed to be well mixed temporally. Typical regulatory monitoring at a sewage treatment plant involves collecting aliquots equally spaced over a 24-hour period (usually one each hour). While appropriate for regulatory purposes, equally spaced composite sampling (in time) may not be adequate for determining “average” concentrations for loading calculations. The user is encouraged to consult up-to-date literature (for example, Ort and others, 2010), when designing a sampling program for sewers and other similar waste streams.

Bottle Types and Configuration

Commercial autosamplers typically hold up to twenty-four 1 L sample bottles, eight 2 L bottles, four 4 L bottles, or a single large carboy of varying volume. Bottles can be made of glass or plastic. Often, equipment specific bottles supplied only by the autosampler manufacture are required. Examples include the plastic wedge-shaped bottles or tall cylindrical shaped glass bottles shown in figure 33. These bottles typically require equipment-specific bottle holders or racks. It is important to use only bottle-holder cages made for a specific autosampler, as any non-specified equipment may result in missed or lost samples. As mentioned above, one alternative is to have a large volume container held outside of the autosampler.

For some bottle types, commercially pre-cleaned sample bottles are available from laboratory supply companies or obtained directly from analytic laboratories. For example, in figure 33, a 4 L glass, class-A level, pre-cleaned bottle is shown that can be used in an autosampler for composite sampling. These bottles are ideal for collecting large volume composite samples for trace-level organic-compound analysis. Similarly, bottles made of perfluorinated plastic (Teflon) are available for trace inorganic metals. Expensive, commercially pre-cleaned bottles are not necessary when sampling for standard inorganic analysis or for analysis of suspended sediment. Whenever glass or plastic bottles are reused, standardized cleaning procedures and blank testing should be followed to assure the cleanliness of the bottles.

Single-Bottle and Multiple-Bottle Schema

The advantage to using a single-bottle autosampler is that a complex sample bottle-distributor system is not needed. Single bottle autosamplers often suffice, for example, when attempting to capture the “first flush” in an intermittently flowing stream or culvert, or to sample when a preset stage is reached, such as at the crest of a storm.

It may be tempting to devise a system where a single composite sample is produced using a system that controls the length of time a peristaltic pump operates. This is not recommended because the efficiency of a peristaltic pump varies with changes in pumping head, and as the pump tubing wears. As a result, equal-volume aliquots will not be obtained, thereby violating a basic requirement for producing a proper composite sample. It is possible to devise a more sophisticated system utilizing a data logger and electronic flow monitor, or incorporate an electronic balance that weighs each aliquot, to produce a single-bottle composite sample.

Multiple-bottle samples, consisting of either discrete or composited samples, can be produced using many of the commercially available autosamplers. These systems utilize a distributor system that places each sample (or aliquot) in an individual bottle. The differences between a single- and multiple-bottle sample are depicted in figure 30. Multiple-bottle composite sampling is used when replicate samples are needed, or when a sufficient volume for the analysis of all components cannot be obtained from a single bottle sample. An example would be a program that samples for sediment-bound trace metals, organic compounds, along with suspended-sediment concentration in a river. To do this, it is necessary to deploy multiple autosamplers set to operate sequentially (from the same triggering signal) to obtain sufficient sample volume to meet the analytical requirements. For simplicity, when deploying multiple autosamplers, use multiple intake lines.

Designing Sampling Systems and Routines for Capturing Storm Events

Flowing water, either in natural riverine settings or in constructed systems (stormwater removal swales, stormwater removal or retention systems, and others) exhibit many different hydrologic responses to precipitation events. These hydrologic responses are also affected by the meteorologic characteristics of the associated precipitation events (for example: short, high-intensity bursts during thunderstorms; long, slowly increasing precipitation as large-frontal systems pass; and rapid snowmelt). It is not the intent of this document to present a detailed discussion of the characteristics of watershed hydraulics during storm events. However, knowledge of the characteristics of the system under study will aid in the design of sampling equipment and control schemes.

When designing a sampling system, it is important to study any available hydrographs of the system or nearby similar systems, along with any available water-quality data; suspended sediment and specific conductance are two of the more important water-quality parameters that may aid in planning a sampling program, but turbidity and major-ion chemistry can also play a vital role (Lewis, 1996; Lewis and Eads, 2001). Other information that is helpful includes detailed maps of the watershed or sewer shed that indicate tributaries, storm drainage pipe outfalls, dams or other impediments, paved areas, and other natural and cultural features.

Hydrographs of storm events are presented throughout this document as examples and, while specific to their watershed, show some of the more general features in hydrographs associated with storm events. These include

There are, of course, a myriad of different combinations of stream response to precipitation events. However, the general nature of the response shown by the system under study and the study goals or regulatory constraints play a role in selecting the sampling equipment and sampling schemes used in a study.

Triggering on Stage or Discharge

Perhaps the most common environmental parameter used for initiating an autosampler is stream stage or discharge. A rapid increase in stream stage associated with a precipitation event is used as a signal to initiate the sampling routine; generally, the autosampler is programmed to collect its first sample (or first composite aliquot) at this point, or it can be instructed to start summing the discharge until the first sampling interval is reached.

As mentioned previously, discharge is a required parameter for producing a flow-weighted composite sample, either by automatic or manual compositing methods. To measure stage, a pressure transducer is typically buried in the streambed or attached to the floor of a culvert, or an acoustic stage or flow monitor is used to initiate sampling. A built-in, or stand-alone data logger, uses a pre-programmed stage-discharge relation to calculate discharge, and then sums the stream discharge to trigger the next sample collection. It is clearly advantageous to locate the auto-sampling site near or at an existing streamgaging station, especially if there is a well-established stage-discharge relation available for the site. For non-gaged rivers, discharge can be obtained from a nearby gaged stream or estimated from basin characteristics until sufficient discharge measurements can be made to construct a stage-discharge relation for the site.

When using stage as a sampling trigger, historic data for the stream can be consulted to aid in programming the autosampler. A general rule is to choose an increase in stage that is at least twice the normal daily variation in base flow, as the trigger point for initiating the sampling program (the collection of the first sample or aliquot). Historic discharge data can then be evaluated to determine the typical variation during base flow.

While it is not possible to precisely predict how a river will respond to an upcoming precipitation event, rainfall data, if available, can be tabulated and combined with historic river discharge data to help in setting initiation stage and discharge intervals. For example, ranges of precipitation totals and intensity were tabulated with corresponding ranges of maximum discharge and duration of storm flow, as function of growing and non-growing season (or other classifications) for two rivers in New Jersey (table 12). This table was compiled using a 25-year dataset; however, using even the previous few years can provide useful information for designing a sampling scheme for an upcoming storm under present-day conditions in the study watershed. A spreadsheet was then developed that, using the historic precipitation or flow data, and the approximate runoff (in cubic feet) expected for a given amount of precipitation, can estimate the number of discrete samples to be collected per day. For composite sampling, the volume of the sample bottle and aliquots can be calculated. This type of spreadsheet, along with weather forecasts, can help researchers choose initial sampling parameters.

When deploying an autosampler on a non-gaged stream, the stage-discharge relation for a nearby gaged stream may be helpful as a starting point for programming discharge-based triggering. However, it is important to ensure that the nearby “surrogate” stream basin is comparable to the studied stream. If data from a surrogate stream are not available, then the researcher should develop a preliminary stage-discharge curve for the study stream by making cross-sectional velocity measurements over a few different flow conditions. The preliminary stage-discharge relation can be refined over time as additional data become available. Flow-composite sampling is not advised at locations lacking an existing stage-discharge relation.

Triggering on Precipitation

Another commonly used sampling trigger is the onset of precipitation or the onset of snow or ice melt. Tipping-bucket rain gages (for example, fig. 34) can be used to initiate autosampler programs, and to collect volumetric measurements. Precipitation sensors that make an electrical connection when sufficiently wetted can be used to initiate an autosampler. These types of precipitation gages and sensors can be deployed throughout a watershed and combined with telemetry equipment to initiate sampling at downstream sites. Care should be taken that the precipitation monitor is deployed using appropriate methods, such as specified by the National Weather Service (National Weather Service, 2024) or other agency.

Triggering Using Other Parameters

Other water-quality parameters can be used to trigger an autosampler program often for specific study goals. A review of chapter A6 of the NFM provides guidance for the standard measuring procedures for these parameters across a range of scenarios and conditions (U.S. Geological Survey, 2008b). This guidance enables study designs in which the following parameters have been used:

Strategies for Triggering Composite Sampling

As described earlier, one use for autosamplers is to automatically produce a flow-weighted composite sample that provides an estimate of an average concentration in a stream over a hydrologic event. Usually this is a precipitation event, but it also can be defined over any range of flow conditions or time. The example in table 13 demonstrates the calculations involved in producing a flow-weighted composite sample from individual discrete samples, or as a single sample produced by compositing equal-volume aliquots collected at equal discharge intervals. In this example, the number of samples, the time intervals, and flow intervals were selected based upon a model, hypothetical hydrograph. For streams that are not gaged, it is often best to start by collecting, and manually compositing, discrete samples. It is also helpful to conduct (and budget for) practice sampling events. This will help the researcher predict how a stream will respond to different storm events.

As an example of how these calculations are used to design a sampling routine, figures 36 and 37 show stream hydrographs for a large river (the Raritan River) and a small urban river (the Elizabeth River) in New Jersey (Wilson and Bonin, 2007). Shown on the hydrographs are times that discrete 1 L samples were collected to analyze suspended-sediment concentration. These samples were collected at equal stream discharge intervals. Additionally at each time point, a second autosampler was used to collect and composite (in a 4 L glass bottle) an aliquot of river water for chemical analysis. Both autosamplers were triggered using the same stage signal.

In the larger Raritan River, two precipitation-fronts had passed through the area within two days. Multipeak hydrographs are common where large-scale cyclonic storm fronts pass through large watersheds that have several large tributaries. However, it should be noted that a similar response may be observed where a tributary enters upstream close to the sampling site. In this example, the autosampler was left operating after the first front had passed through the area. Then the larger precipitation front passed through the basin, resulting in the second much larger portion of the storm hydrograph (fig. 36). The concentrations of suspended sediment in the figure are connected using stepwise interpolation, then integrated to determine the suspended-sediment load. As can be seen, the suspended-sediment concentrations in the stream peaked early, but after the second peak, SSC decreased and remained low despite the high discharge.

In the smaller Elizabeth River, a short, high intensity rain event had resulted from a summer thunderstorm. This produced three “flashy” discharge peaks; these peaks result from discharge emanating either from nearby upstream or storm drains (fig. 37). In this case the suspended-sediment concentration reached a maximum in the first discharge peak, and lower peaks of concentration were measured in subsequent peaks of discharge. This phenomenon is likely the result of “sediment depletion” as the river and tributaries become flushed of easily moved fine-grained sediment that was stored in the river since the time the most recent storm event occurred. The discharge interval used in this sampling allowed for an accurate picture to be developed of how the sediment concentrations varied with time. However, the spacing was large enough to keep the discrete samples from overlapping during maximum discharge.

Designing a Program for Automated Composite Sampling of a Storm Event

Reviewing historic discharge and precipitation data for an area is the first step in designing a successful sampling scheme. The unknown variables are the amount and intensity of precipitation in the upcoming storm event, the appropriate discharge interval for collecting each sample, and the aliquot volume that will produce an accurate composite sample. The following are some strategies that may help in choosing these variables:

Autosampler Maintenance and Cleaning

All autosamplers and data loggers will require servicing before each sampling event. The servicing should include general cleaning, verification that adequate battery or electrical service is available, examination (and sometimes replacement) of pump tubing and inlet lines, and restocking of clean bottles. A checklist should be provided for these steps.

Proper cleaning of all the components of the autosampler that come in contact with a sample is essential to ensuring valid results. Verification of equipment cleanliness and due-diligence by the field crews is provided by the field-equipment blanks and other QA/QC samples. Depending on the type of autosampler used, the parts that may need cleaning include the intake line, pump tubing, diversion valve (if used), distributor line tubing, sample container(s), and interior of the autosampler housing. As a general rule, the cleaning instructions for equipment used are based on the constituent(s) under study. In addition, each autosampler may have specific manufacturer’s instructions for cleaning, and should be followed.

Different classes of analytes require special cleaning needs. For general cleaning instructions for inorganic and organic sampling see chapter A3 of the NFM, “Cleaning of Equipment for Water Sampling” (U.S. Geological Survey, 2004). Table 14 summarizes the general cleaning requirements for major parameter groups. See also the “Cleaning Procedures” section of appendix 2.

Intake-Line Maintenance and Cleaning

The intake line is one of the most vital pieces of the sampling system and has great potential to affect the integrity of samples. Deposits of sediment, organic matter, oils, and grease can affect the chemistry of each volume of water passing between a stream and sample container. For example, figure 18 shows how inlet lines can become coated when left in a river for extended periods of time. Therefore, it is important that a rigorous and thorough cleaning program is followed, especially when autosamplers are deployed in a long-term or permanent study and for trace components. Intake lines and screens (if used) should be removed, checked, cleaned, or replaced before each sampling event following steps documented in the study SOP. As a general policy, it is advisable, at a minimum, to clean the intake lines used to collect chemical samples before each event. Replacing lines with pre-cleaned tubing for each event should be considered and may in fact be necessary when sampling urban streams or waste outfalls carrying water containing oils and grease. The ability to replace intake lines should be considered carefully when designing a sampling system, especially where trace elements or organic compounds are being studied. Prudent operating procedures dictate, however, that intake lines are pulled for cleaning or replacement on at least a regular basis.

Lines used to collect only suspended sediment do not necessarily need to be replaced before each event but should be inspected periodically to determine if sediment is accumulating in low points (dips). Sediment will accumulate if pumping rates are insufficient, or if lines are left unused for extended periods of time. Pumping tap water or air under high-pressure may be used to purge this sediment from the line.

It is important to inspect the inlet line where it enters the stream, along with the nozzle and screen if used, to verify it is not kinked, obstructed, or excessively abraded. A small portion of the nozzle can often be cut to remove a kink or rough surfaces, but the cut should be sharp and clean. Screens should be removed and cleaned with hot-soapy water before each event or replaced with a pre-cleaned screen. Pump and distributor arm tubing should be replaced with new pre-cleaned tubing before each event.

Cleaning is best performed in the laboratory but in some instances (for example, lines used for suspended sediment or general inorganics) can be conducted in the field if necessary.

In the laboratory, the various inlet and pump tubes can be connected together using a plastic barbed fitting and attached to form a single line. Using a peristaltic pump, the sequence of appropriate cleaning solutions (described earlier for the different intended constituents of interest in table 14) can be pumped through the entire length. For safety, it is recommended that cleaning solutions be “pulled” through the lines, in case a break in a connection occurs. The following is a general sequence of cleaning steps:

Use a phosphate-free laboratory detergent solution no stronger than 0.2-percent (2 mL concentrated detergent per liter of water); stronger solutions do not perform better, and even a small amount of residual can be carried over or leave a residue on tubing and fittings. Although the acid and methanol rinses are not required when the lines are used for general inorganic constituents, nutrients, or sediment, they are necessary when sampling for trace metals and organic constituents and are recommended in all studies as these solutions will remove any oil or grease adhering to the inside of the previously used sampling lines. Allow the cleaning solutions to recirculate continuously for 20–30 minutes before disposal following laboratory procedures. Periodically stop the pump and let the solutions stand for several minutes in the tubing. While the solutions are circulating, periodically pinch the tubing outlet to allow the solution and air bubbles to mix. Once cleaned, the ends of the inlet lines should be capped with foil or plastic wrap, the tubing carefully wound, and placed in large plastic bag for transport.

Cleaning inlet lines in the field in situations that require removing and replacing the lines is difficult. As mentioned previously, to remove any debris that may have accumulated in the lines, it is advantageous to occasionally purge the lines with high pressure air or tap water before cleaning. The cleaning solutions can then be pumped down the lines (from the site into the stream) using a peristaltic pump or the autosampler. A waste container should be placed at the nozzle end of the line to collect solutions. Stop the pump occasionally for 5–10 minutes to allow the cleaning solutions (soap, acid solution, or methanol) to remain in the lines and soak. Try not to allow acid, soapy water, or methanol to discharge into a natural water body. Note that field cleaning is most practical for locations where large volumes (25–50 L) of tap or distilled water and necessary cleaning solutions can be easily transported to the station.

Following retrieval of samples and sampling data, the autosamplers will need servicing before the next sampling event. If events are to be separated by weeks or months, it is best to remove equipment and sampling lines. Table 15 lists many of the steps recommended when servicing an autosampler; these steps should be detailed in the project SOP. All of these steps are important; however, some are more critical to ensuring the success of subsequent sampling efforts. For example, it is critical that the sample volume is calibrated each time the autosampler is serviced, especially if composite samples are being produced, and whenever pump tubing is replaced. Follow the manufacturer’s instructions for confirming and adjusting the sample volume. Field crews should be provided step-by-step instructions for calibrating and documenting pump preparations. Where a stand-alone data logger is used to control pump equipment, the operating code should provide the capability for calibrating pumps.

Chapter 4 of the National Field Manual (U.S. Geological Survey, 2006) and Britton and Greeson (1989) provide helpful guidance on the general sequence for cleaning equipment and bottles for sampling inorganic or organic analytes. However, depending upon the analyte suite (trace metals, organics, and bacteria for example), additional cleaning and preparation steps may be required.

Retrieval of Samples and Data

Regardless of the purpose for which the autosampler is deployed, it is prudent to describe in the standard operating procedure (SOP) document how samples and data are to be retrieved, and how equipment is to be maintained. Table 16 lists some of the steps that may be involved in servicing an autosampler. The SOP should outline all of the steps to be followed, provide forms that help field personnel inventory and record the necessary data, and steps to verify the autosampler was properly calibrated and programmed for the next sampling event (if applicable). It should list how the data are to be downloaded from the autosampler (or data logger), and the information that should accompany samples to the analytic laboratory. An example SOP is provided in appendix 2.

Sample Processing

Samples should be retrieved as soon as possible upon the conclusion of a sampling event. Some constituents such as bacteria or VOCs have short, required holding times (if applicable), and the sampling plan should dictate the retrieval and preparation of the samples as soon as possible, perhaps even when the autosampler is still operating. The first step in retrieval should be to ensure all samples are secured, properly capped and labeled. At a minimum, the label should contain date and time of collection or retrieval, and the location of the bottle in the autosampler. These data can then be recorded on field forms. When discrete bottles are being retrieved, it is advantageous to maintain the bottles in their respective place in the autosampler carousel, and number their position directly on the bottle in permanent marker. Be very careful not to switch bottle positions in the carousel. Note: If the autosampler is being readied for the next event, replace the carousel with a new carousel containing unused bottles.

Other general guidelines include the following:

Sample Documentation and Metadata

Metadata are the ancillary information necessary to make samples useful for study purposes. It is imperative that all needed metadata be recorded upon retrieval of samples for a sampling event. The SOP should contain a check list detailing all metadata required for each sample, along with steps for processing and labeling samples, the protocols for collecting QA/QC samples, a detailed schedule for servicing, and information on programming the autosampler. Table 7, table 17, and table 18 list some of the typical metadata that may be needed during deployment of an autosampler.

Summary—Autosampler Deployment