A systematic approach to choosing an automated nutrient analyser for river monitoring

Abstract

Automated high frequency nutrient analysers have recently become available for in-stream monitoring of freshwater ecosystems. These instruments permit observation of nutrients at the same temporal frequency as discharge measurements. In principle this development will overcome some of the limitations of current water quality sampling and enable a better understanding of coupled terrestrial and aquatic environmental systems. This paper presents a systematic approach to choosing such instruments for research applications and informing the design of prescribed water quality monitoring. The instruments considered are ion-selective electrodes, wet chemistry analysers and ultraviolet/visible light spectrophotometers. Before committing to a new technology, investigators should evaluate instrument related considerations and complementary, often project-specific factors, in a structured way. The instrument related considerations are the ability of the instrument to measure the required nutrient parameters, the temporal resolution, the detection limits and range of individual measurements, required accuracy and operating temperatures as well as the overall cost. The complementary factors to consider are the maintenance effort, operating conditions, major service expenses and special consideration associated with individual instruments. This evaluation is presented for a range of available instruments across the three instrument types. As supplementary material a tabular approach that combines these factors is proposed and illustrated with a case study where instruments were selected for researching nutrient movement in a catchment in northern Tasmania, Australia. Few of the instruments can provide all the essential requirements of the case study and significant compromise of maintenance costs and functionality was necessary. The approach is readily adaptable to choices of instruments for a wide range of investigations concerning aquatic water quality. Clearly the outcome of the choice process is likely to be different for different applications, locations and environments.

---

Environmental impact

High frequency nutrient instruments have recently became available for in-stream monitoring of freshwater ecosystems. In principle these new technologies permit the observation of nutrients at the same temporal frequency as discharge measurements. This paper gives an overview of these new technologies and sets out a suggested process of instrument selection for a range of applications.

In principle the application of these instruments will overcome some of the limitations of current water quality sampling and enable a better understanding of coupled terrestrial and aquatic environmental systems. In practice the instrument-specific detection limits, maintenance requirements, cost considerations, maximum sampling frequencies and the number of constituents measured, all limit the scope of their use.

1. Introduction

Water quality observations are fundamental to the analysis of water resources and water health responses.^1–3 For aquatic environments elevated nutrient concentrations remain as a major cause of water management concerns.^4,5 Human induced increases in nutrient concentrations have been linked to changes in food chains, episodic hypoxic or anoxic and temporary acidic events, macro invertebrate and fish kills in water systems.^6,7 The importance of water quality monitoring has been recognised world wide by legislation and the subsequent implementation of basin scale federal programs such as the USGS National Water Quality Assessment (NAWQA) Program and the US EPA Total Maximum Daily Load Program (TMDL),⁸ the European Water Framework Directive⁹ or the Australian National Water Quality Management Strategy.¹⁰

Until recently practical considerations concerning the number of samples to be taken and the field service resources have restricted the frequency of water quality samples taken at an observational station¹¹ for research and prescribed monitoring purposes. Field and laboratory costs and error accumulation in analysis are primary concerns in the design of existing sampling.¹² Autosamplers have increased the timing and number of measurements but are limited by small sampling capacities and preservation issues in unfiltered samples.^13–15

A new generation of instruments for measuring in-stream nutrient concentrations has now become available. These instruments are field-deployable water quality analysers suitable for freshwater systems and are marketed as overcoming the drawbacks of current manual water quality sampling strategies. Much of this technical work has been summarised in three monographs by Buffle and Horvai,¹⁶ Varney,¹⁷ Down and Lehr¹⁸ and papers e.g. by Johnson et al.¹⁹ and Muller et al.²⁰ In recent years a small number of research studies that utilise these instruments operating at a high temporal frequency have been published in the hydrologic, biogeochemical and limnology literature.^13,21–23

Decisions concerning investments in instruments and the associated maintenance are complex and concern many application-specific factors. There are a number of motivations driving the rapid evolution of instruments for measurement of river water quality. Addressing the costs associated with “sample, then store, then analyse in the laboratory” approaches is a significant driver. New systems can address these costs by removing the storage step and moving the analysis instrument into the field. This approach has the further advantage of reducing or eliminating concerns that the sample may change during its storage and transport. A further motivation is the world wide trend to real-time or near-real-time reporting. New systems with field deployed analysis offer the prospect of water quality being reported via telemetry, web-services or other online data protocol technologies. This development in turn can enable alert services where measurements above a pre-set threshold can trigger intervention.

A further group of advantages of field deployable instruments compared with laboratory-based measurement is the increased ability to use high frequency measurements: 1. the calculation of nutrient fluxes is much improved as the estimate of representative concentrations is less subject to sampling errors (e.g.ref. 24), 2. alerts at trigger concentration values will also be improved due to the reduction in sampling errors (e.g.ref. 19,25), and 3. high frequency measurements are suitable for tracing the full temporal scale of various nutrient sources in catchments.^21,26,27

The choice of an automated high frequency analyser depends on the capability of the various instruments, the intended application and the operating environment. This paper provides a structured approach to assessing the many issues associated with field-deployable water quality analysers. The reader is provided with a description of the considerations on each issue and then a method of synthesising these factors into a single choice process. While the illustrative example is in the research and development domain, the choice method will also have some application in prescribed monitoring applications.

2. Field deployable analysers

Currently there are three different types of field deployable water quality instruments available that have the potential to measure the concentrations of pollutants in freshwater systems: non-spectroscopic instruments like ion-selective electrodes, wet chemistry analysers, and spectroscopic analysers including optical UV/Vis systems. Here we provide a brief description of the measurement principles. In the following sections operational considerations for each type of instrument will be described.

2.1 Ion selective electrode (ISE)

Ion-selective electrodes utilise a potentiometric method.^18,28 Using an ion-specific membrane targeted ions are selected and measured through comparing the induced electrical potential to a reference potential.^18,28 Ion-selective electrodes are currently available for a range of pollutants including the nutrient species nitrate and ammonium. The standard field-deployable ion-selective electrodes consist of a protective housing storing the measuring electrode immersed in a reference solution of the analyte ion, the membrane that separates the chamber for the sample solution, as well as a reference electrode, potentiometer and power source. However there are many more different types of ion-selective electrodes available depending on the membranes and microelectrodes used.²⁹ Ion-selective electrodes can be immersed or built as parts of mini analysers. In practise it is important that ISEs are compensated for interfering ions, such as Cl for NO₃ ions and K for NH₄, for example through adding complexing agents or compensation membranes. According to^19,30 ion-selective electrodes are not widely used for autonomous unattended monitoring applications in practise due to the need for manually re-calibrating every few hours.¹³ However, new generations of ion-selective electrodes membranes might be stable for a longer period without re-calibration.^29,31–33

2.2 Wet chemistry analyser

Wet chemistry analysers are based on standard spectrophotometric methods originally developed for laboratory purposes. A variety of standard colorimetric methods for measuring nitrogen (N) and phosphorus (P) species have being adapted for field deployment.¹⁹ These different methods are summarised for example by Gray et al.³⁴ During the analysis the nutrients are in principle reduced by chemical or photo-reduction, followed by a colour development (pink azo dyes or molybdate blue dyes) and then measured at a specific wavelength within a colorimeter. The measurement of total N and P involves a digestion step, which converts organic and particulate N and P into mineral forms. This is usually carried out by UV or microwave methods.³⁴ As a consequence total N and P analysis require more time to measure than their dissolved species.

The field deployable on-line measurements possible include dissolved nutrient species like nitrate, nitrite, ammonia and orthophosphate as well as total nitrogen and phosphorus among other photometric detectable ions. Hardware components include pumps for sample intake and injecting reagents, tubing, valves, a manifold for reagent addition and a colorimetric detector. These components are mounted in a variety of ways,³⁴ so that they operate as flow injection analyser (FIA), sequential flow analyser (SFA), or segmented continuous flow analyser (SCFA). The wet chemistry instruments are either designed as modular mini-lab systems with a choice of ions that can be measured, or as in situ analysers with more limited amounts of measurements. In order to correctly distinguish dissolved nutrient components a continuous 0.45 μm filtration is needed.

Wet chemistry analysers require chemical reagents. Since reagents deteriorate over time auto-recalibration is essential to improve accuracy and data quality. In practice, interferences with other constituents can occur. For nitrate analysis the reduction efficiency of nitrate to nitrite can vary over time as well as the methods used for the reduction phase vary in their efficiency.³⁴

2.3 Optical UV/Vis sensors

UV/Vis spectroscopy is based on the selective adsorption of electromagnetic radiation in the ultraviolet and visible wavelength region. Currently this method is limited to the analysis of nitrate, though other constituents like organic solids, turbidity or pesticides are possible.^35,36 These instruments measure the raw absorbance spectrum of water^14,37 at multiple wavelengths. For nitrate the measured range is in the UV spectrum (180–370 nm). For some instruments the absorption range is extended (up to 780 nm) in order to determine additional constituents, like turbidity, TOC and DOC by using specific deconvolution algorithms. There are also instrument differences in UV/Vis wavelength resolution, with some instruments measuring absorbance at a 1 nm wavelength interval and others at a 2 nm interval. The solvent-free, field deployable instrument consists of a protective housing cylinder including a UV/Vis light source, fibre optics, and spectrometer, battery or power access cable. Real time data deconvolution and instrument control is achieved by coupling the sonde to a microprocessor. Deconvolution algorithms are undergoing development to reduce known interferences.³⁸

The measurement of the absorbance spectrum and of the deconvolution algorithm is very rapid and some instruments are scalable for a variety of nutrient concentration sensitivities. Designed as in situ instruments UV/Vis analysers can be deployed on-site either submerged or with pumped sample.

3. Considerations for instrument selection

In this section we present a structured summary of considerations for the instrument types listed above. The factors to be taken into account are divided into ‘instrument related considerations’, which are described individually, and ‘complementary factors’ that are presented as a group. Instrument related considerations are factors associated with the instruments, whereas complementary factors need to be considered in relation to the proposed applications.

Table 1 provides an overview of the essential considerations described below and was collated from nine commercially available systems. The authors have made their best efforts to describe current specifications and prices accurately but these are subject to change.

Table 1 Specifications for selected high frequency nutrient analysers (1–9) for freshwater systems with special emphasis on the detection of low nutrient concentrations

	ISE		Wet Chemistry					UV/Vis
Provider/Name of equipment	(1) WTW(™)/VARiON	(2) Greenspan (™)/Aqualab	(3) YSI(™)/YSI96000	(4) Systea(™)/Micromac C	(5) Ecotech(™)/FIA NUT1000	(6) EnviroTech/(™) AutoLAB or MicroLAB	(7) FIALab(™)/SIA	(8) Satlantic(™)/ISUS	(9) S::can(™)/Spectronalyser
a Indicative price of instrument only in $AUS (please note that prices may reflect vendor pricing policy to different customers, readers are encouraged to check for current prices themselves. b Detection range to be fully defined by the customer, dilution factor used in range of twice of the upper defined detection limit. c Weight in air without reagent. d Not applicable. e Measuring range (wavelength, nm).
Monitored parameters	NH4–N, NO3–N	NO3–N (ISE), NH4–N (ISE), PO4–P (wet chemistry)	NO3–N	TN, NO3–N, NO2–N, NH4–N, TP, PO4–P	PO4–P	NO3–N, NH4–N, PO4–P	NO3–N, NO2–N, NH4–N, PO4–P	NO3–N	NO3–N + NO2–N,
Capital cost/unit^a	$ 7500 (NH4 and NO3)	$ 68 000	$ 29 250	$ 95 000	$ 18 000	$ 46 500	$ 100 000	$ 54 600	$ 45 000
Minimum temporal resolution	20 s	15 min (single channel), 50 min (multi channel)	30 min	15 min (single channel), 38 min (multi channel), 60 min for TN & TP (multi channel)	30 s (1/30 Hz)	15 min (single channel), 60 min (multi channel)	15 min	30 s (1.0/30 Hz)	30 s
Detection limits and range [in mg L^−1]	NO3: 0.1–100	NO3: 0.1–14	NO3–N: 0.005−2	TN, NO3–N: 0.002	PO4–P: 0.002–0.120	NO3–N: 0.002–13	NO3–N: 0.003	NO3–N+ NO2–N: 0.007–28	NO3–N + NO2–N: 0.005–7
	NH4: 0.1–100	PO4: 0.1–3		NO2–N: 0.002		NH4–N: 0.002–4	NO2–N: 0.001
		NH4:0.3–14		NH4–N: 0.002		PO4–P: 0.002–1.5	NH4–N: 0.010
				TP, PO4–P: 0.001			PO4–P: 0.003
				Customizable^b dilution factor			Customizable^b
Ambient operating conditions	0–40 °C pH dependence	5–50 °C	1–45 °C	NO3–N, NH4–N, PO4–P: 4–40 °C, TN & TP: 10–30 °C	n.a.^d	n.a.^d	n.a.^d	0–45 °C	0–45 °C
Claimed accuracy	± 5% of range	± 5% of range (NO3 & NH4), ± 2% of range (PO4)	± 5% of range	± 3% of customizable range	± 2% of range	± 2% of range (NO3 & NH4), ± 3% range (PO4–P)	± 2% of range	± 2% of range	± 3% of range
Telemetry	yes	yes	yes	yes	yes	yes	yes	yes	yes
Warranty	2 years	1 year	1 year	2 years	1 year	2 years	1 year	1 year	2 years
Additional information	Interference compensation	n.a.^d	In situ weight 18.2 kg ^c	In situ weight 36 kg ^c	n.a ^d	In situ weight 25 kg ^b	n.a ^d	180–370 nm^e Built-in filter	180–780 nm^e Self cleaning unit

---

3.1 Instrument related considerations

3.1.1 Monitored nutrient species and frequency. Wet chemistry analysers are able to measure the greatest range of N and P species, but individual brands differ in their capabilities. UV/Vis and ISE can only measure a limited number of analytes but they offer the greatest potential for fast measurement, with programmable frequencies down to 20 s. In contrast wet chemistry analysers need about 15 min for each analyte, increasing if more constituents are determined for the same sample. The sampling interval for total nutrient concentrations (TN and TP) is greater (up to 1 h). All analysers are programmable for greater time intervals.

3.1.2 Detection limits and range. Wet chemistry analysers obtain the lowest nutrient concentrations at around 0.002 mg L⁻¹. Although considerable effort is going in the development of low sensitive ion-selective electrodes,^29,33,39 in practise we found no field-deployable ISE with detection limits less than about 0.1 mg L⁻¹. The measurable concentration range for each instrument is a critical factor as they need to match the proposed sampling environment. The detectable concentration range is either fixed or can be customised. Some wet chemistry analysers are capable of extending the upper detection limits by incorporating a dilution factor. In practise the importance of the detection limits depends on the purpose of the proposed monitoring strategy. It is important to choose whether the entire anticipated concentration range should be measurable or if the monitoring is used to observe concentrations above a certain thresholds (e.g. compliance monitoring). Clearly, more instruments will be suitable for compliance monitoring than for applications where the widest operating range and lowest detection limits are desirable.

3.1.3 Cost considerations. In the selection process it is important to carefully calculate instrument related costs. Capital costs vary significantly for the instruments considered (Table 1). Wet chemistry analysers tend to be more expensive, but prices also depend on which measurement capabilities are chosen. Ion-selective electrodes are the cheapest instruments. The costs of UV/Vis instruments are in the order of a wet chemistry analyser with capability to measure two–three constituents. But the wet chemistry analysers incur additional cost for chemical reagents: purchase, preparation and/or possibly for its waste disposal. These costs need to be carefully calculated as they might exceed the capital costs. For some instruments supplemental acquiring costs need to be factored in. This includes additional hardware components, like instrument or microprocessor controllers, a data logger, telemetry or recommended components like on-board cleaning units.

3.1.4 Ambient operating conditions. Care should be taken to assess whether local environmental conditions fall within the specified operating range for the instrument. The specified temperature range for ISE and UV/Vis (from about 0 °C to 40 °C or 45 °C) is slightly wider than for wet chemistry analysers. However, this comparison is complicated by temperature impacts on reagent stability for wet chemistry analysers. Rapid temperature changes accelerate the decay of some reagents, even within the recommended range. Our experience was that some highly concentrated reagent crystallised within the supplier stated temperature range (e.g. Tris-Hydroxymethylaminomethane, Orthophtalaldehyde) causing instrument failure. This may have been due to the instrument being exposed to prolonged periods of temperature near the lower limit. Alternatively it may have been due to rapid temperature decreases during the night. Also some reagents are sensitive to light and therefore need to be stored in a dark environment. Other reagents (e.g. sodium hydroxide) are prone to contamination and require special handling (e.g. CO₂ traps). The environmental condition limitations described here are likely to be a significant constraint in many applications. These constraints can generally be removed through the installation and maintenance of climate-controlled instrument housing. An operating limitation for the ISE instruments exists outside certain pH ranges (pH 4–8.5 for NH₄ and pH 4–11 for NO₃) but those pH ranges are unlikely to be exceeded in most freshwater systems.

3.1.5 Claimed accuracy of measurements. The accuracy measures the degree of closeness of a measured to its actual concentration value. For all instruments the measuring accuracy is claimed by the manufacture of the instrument to be 5% or less of the full scale. The potential error can therefore be large when using an instrument in flows with low concentrations and wide detection ranges. In practise the accuracy is lower at the detection boundaries. It is beyond the scope of the paper to independently check the claimed accuracy of each of the instruments. However, as per normal practise, checks for a selected instrument should be carried out for the range of nutrient concentrations that are likely to occur. Costs associated with the testing should be factored in.

3.1.6 Telemetry and warranty. All considered analysers can be equipped for telemetry and alarm setting, but technical limitations such as transmission range might apply. Some instruments can be set up for remote calibration. Data transmission can be realised via mobile phone links, land lines or satellite phone. Associated costs vary greatly depending on the amount of data transfer and local transmission rates.

Warranty is typically given for between one or two years. However specification of the individual hardware components covered by the warranty is advantageous.

3.2 Additional factors

Complementary considerations cover factors that are related to a proposed application and are summarised according to instrument types. This includes the maintenance effort, durability, installation and housing as well as power, data management, and other support. Some factors differ significantly between the three instrument types. Table 2 provides a comparison of these.

Table 2 Complementary considerations in selecting type of instrument for high frequency water quality monitoring

	ISE	Wet Chemistry	UV/Vis
Maintenance effort	Recalibration and cleaning	Reagent replacement and cleaning	Cleaning
	Depends on manufacture: Ranging from 1 day to 6–8 weeks	Exchange frequency depends on reagent consumption and reagent stability	Depends on local build-up rate of algae or other material
		Ranging from 2–8 weeks	Ranging from 2–8 weeks
Installation and housing	In situ (mostly) or off site	In situ or off site (mostly)	In situ (mostly) or off site
	Instream securing from floods, debris and cobbles	Instream securing from floods, debris and cobbles	Instream securing from floods, debris and cobbles
	Removal during periods of river icing	Off-site securing: insulation for enclosure and risk management for chemicals	Removal during periods of river icing
		Reagents might be sensitive to light and rapid temperature changes causing instrument failure
Major service issues	Membrane decay	Hardware components	UV light
	Depends on type of instrument	Depends on instrument usage and quality of hardware components	For ISUS: UV light needs exchange after 1000 h of operation
	Ranging from 6–18 months	Annual overhaul for exchanging tubes, filters and peristaltic pump
Special considerations	Additional costs:	Additional costs:	Additional costs:
	Self cleaning unit (∼ $ 2500)	Reagent and standard preparation (∼ $500–$7500 per year)	Self cleaning unit (∼ $1000)
	Membrane exchange (∼ $ 2500 per year)	Waste disposal (∼ $500 per year)	Controlling terminal (∼ $10 000)
		River Pump	Mounting material
	Controlling terminal (∼ $ 3500)	Mounting material	For ISUS: UV light exchange (∼ $2500)
	Mounting material	Filtration	Replacement at factory taking ∼ 4 weeks for servicing
		Considerations for hazardous reagents:	Considerations for operation:
		Implementation of health & safety plans for preparation, packing, transport, storage and waste management handling	Possible need to adapt for sensitivity due to changing river conditions (∼ $2000 for an insert to shorten instrument pathlength)
		Prevention of reagent contamination (e.g. CO2 traps)

---

The required maintenance effort should be critically appraised as it can involve significant cost. Table 2 provides an indication of the maintenance tasks required for each instrument and an estimate of the time required and its frequency. This estimate is based on advice from technicians working for the vendors and our independent experience. Ion-selective electrodes can require frequent re-calibration. Some ISE instruments need manual re-calibration during field visits whereas others are equipped with auto-recalibration routines. The re-calibration frequency depends on the type of membrane used and varies between a day and several months.

Ion-selective electrodes and UV/Vis instruments can be deployed directly in the water body or in a pumped line. If deployed directly in the river then instrument cleaning to mitigate the inevitable biofouling is likely to be a controlling factor on the frequency of maintenance visits. Automated cleaning features like mechanical wipers, and air-pressure or chemical cleaning systems can increase the inter-maintenance period.

Securing and protection requirements differ for in situ installations and those that are deployed on-site, e.g. on the river bank. In situ installations require protecting the instrument against floods and water-borne debris or boulders. During periods of frost the instrument may need to be removed. On-site installations require protection for unwanted biological and hydro-climatic threats. For on-site wet chemistry analysers special care must be taken for reagent risk management (e.g. prevention of reagent spills) and maintaining a suitable instrument operating temperature.

Of the instruments evaluated, wet chemistry analysers have the highest maintenance requirement. Care must be taken at all steps including reagent preparation, reagent replacement, cleaning and waste management. Reagent preparation should be conducted by a certified laboratory to ensure reliable standards. Some of the reagents are hazardous therefore handling and transport of reagents, and waste disposal needs to be carried out carefully according to health and safety guidelines. The required volume of reagents and deionised water can be large especially when monitoring multiple nutrients at high temporal frequency.

A further consideration is costs associated with periodic major instrument services. For ISE instruments membranes need to be exchanged every 6 to 18 months. For UV/Vis instruments there are comparable costs as one particular instrument specifies replacing the UV light on a regular basis. Wet chemistry analysers require at least an annual overhaul in order to replace hardware components (peristaltic pumps, tubing, filters, valves). Their durability depends on the quality of incorporated hardware components and accessories (e.g. UV lamps, river pump).

All instruments have considerable power demands for various combinations of pumping and cleaning and, in the case of the wet chemistry analyser, heating for analysis steps. The power requirements could be met by extensive solar or wind generation installation but the cost may be prohibitive. In our case, the use of mains power was cost-effective and minimised the risk of loss of data from power outages. For all instruments the essential aspects of telemetry are good mobile telephone reception, a transmitting bandwidth that matches the local reception arrangements and specialist support from the equipment supplier.

All high frequency instruments will provide a large amount of data. It is advisable to plan and implement a database management procedure for storing, documenting and archiving incoming data. Data needs to be checked and documented for data quality. This includes controls against obvious outliers (e.g. measured values above and below detection limits), for missing values (e.g. through instrument failure), drifts and interference with other ions (e.g. removal of spikes). Commercially and freely available software products can aid these procedures.

4. Instrument selection: a case study

To make a choice of instrument for a particular application the factors described above should be considered objectively. Here we provide an approach to this problem and illustrate it with our experience as part of a research project in northern Tasmania (Table 3). The details of this case study can be found in the supplementary material.‡ This case study included investigation of hydrologic and biogeochemical process controls and the influence of land use on nutrient delivery within a 369 km² catchment. Water quality analysis instruments were chosen to measure the concentrations of various nitrogen and phosphorus species at a high temporal frequency and for a long term deployment (>3 years) in specific areas of the catchment.

The logic of the key factors in the case study (Table 3) and the illustrated choice is provided in the supplementary material.‡

Table 3 Combining essential and supplementary factors for case study instrument selection^ab

	ISE		Wet Chemistry					UV/Vis
a ++ = highly acceptable, + = acceptable, o = partly acceptable, − = not acceptable, * = factors influential for decision making process. b Numbers refer to instruments collated in Table 1.
Considering essential and supplementary factors	1	2	3	4	5	6	7	8	9	Comments for decision making
Project specific assumption	No submersible wet chemistry analyser considered									As limited reagent capacity
	No limitation for power supply									Power available on site
	No limitation for telemetry									Available for all systems
(1) Instrument related factors
Number of constituents	+	+	o	++*	o	+	+	o	o	3, 5, 8, 9 limited in monitoring constituents
										4 all dissolved species and TN & TP possible
Detection limits	−	−	+	++*	+	++*	+	+	+	1, 2 excluded due to inability to detect small concentrations
Range	−	−	+	++*	o	+(+)*	+	+	++*	1, 2 see above, 5 range is too limited, 4 customizable range and in-built dilution factor, 6 range appropriate solely for NH4 and PO4, 9 adaptable for various concentration ranges
Temporal resolution	+	+	+	+	++	+	+	++	++*	All instruments have sufficient temporal resolution 5, 8, 9 enable very rapid frequency
Claimed precision	o	++	+	++	++	++	++	++	++	Sufficient for all application, but 1 is prone for a higher absolute error due to wider concentration range
(2) Financial constraints
Capital cost per unit	++	+	+	+/o	+	+	+/o	+	+	4, 7 are the most costly systems, but are in the limits of the project budget
Maintenance cost	++	+	+	+	+	+	+	++	++*	1, 8, 9 comparable low, for 2, 3, 4, 5, 6, 7 doable, but need suitable source for reagent preparation, maintenance and waste management
(3) Operating conditions
Ambient temperature	+	+	+/o	+/o	+/o	+/o	+/o	+	+	Feasible for all analysers, but for 3–7 reagent stability is critical in relation to environmental temperature conditions
Installation and housing	+	+	o	+	+	+	+	+	+	Feasible for all analysers but for 4, 5, 6 or 7 protective housing is required, less feasible for 3 as originally designed as submersible
Unattended operation	+/o	+	+/o	+/o	+/o	+/o	+/o	+	++*	9 favourable with automated cleaning, possible for 3–7 assuming reagent temperature control for extreme operating condition
(4) Additional considerations
Durability	+	+	+	+	+	+	+	++	++	8 & 9 less prone to errors due to their rugged construction
Warranty	++	+	+	++*	++	++	+	+	++*	4, 6 and 9 offer a warranty of 2 years
Decision				√					√

---

5. Conclusions

Automated high frequency nutrient analysers have recently become available for monitoring freshwater ecosystems via in-stream sampling and on site analysis. It is tempting for researchers to commit to these new measurement approaches given the claimed advantages. This paper has provided a structured approach to selecting a fit-for-purpose field-deployable water quality analyser. This simple tabular approach was instructive in instrument selection as demonstrated using a research case study from northern Tasmania. It is readily adaptable to other applications. It is recommended that the user of this method be cautious about supplier claims on instrument performance and consider independent experience in the instrument selection.

While the findings in this case should not be over generalised, it can be observed that wet chemistry analysers are more suitable to applications where a wide range of nutrient species are to be measured and required detection limits are low. However, these instruments are complex and currently have relatively high maintenance costs. UV/Vis instruments appear to be suitable for application where nitrate is of importance and there is an emphasis on high frequency measurements. Ion-selective electrode instruments may be appropriate for studies with higher detection limits, frequent site visits and smaller budgets.

The case study evaluation presented here will need to be updated as new instruments come on to the market. New sensors such as biosensors, photothermal analysers or MEMS^5,25 may overcome some of the limitations associated with the current high frequency nutrient analysers.

Acknowledgements

This study was funded by the Australian Commonwealth Environmental Research Facilities programme (CERF) through the Australian Department of the Environment, Water, Heritage and the Arts (DEWHA). The authors acknowledge the cooperation of selected vendors who provided detailed instrument specifications. We thank Hamish Cresswell and Warren Hicks, CLW Canberra and the reviewers whose comments helped to improve the quality of the paper. We also acknowledge the practical guidance concerning the illustrative example given by Hamish Cresswell, Chris Drury, Danny Hunt, Seija Tuomi and Kirsten Verburg.

References

1. M. Meybeck, The global change of continental aquatic systems: dominant impacts of human activities, Water Sci. Technol., 2004, 49(7), 73–83.
2. A. Ducharne, Importance of stream temperature to climate change impact on water quality, Hydrol. Earth Syst. Sci., 2008, 12(3), 797–810.
3. R. B. Jackson, et al., Water in a changing world, Ecol. Appl., 2001, 11(4), 1027–1045.
4. N. E. Peters and M. Meybeck, Water quality degradation effects on freshwater availability: Impacts to human activities, Water Int., 2000, 25(2), 185–193.
5. J. Dutt and J. Davis, Current strategies in nitrite detection and their application to field analysis, J. Environ. Monit., 2002, 4(3), 465–471.
6. R. E. Turner, N. N. Rabalais and D. Justic, Gulf of Mexico hypoxia: Alternate states and a legacy, Environ. Sci. Technol., 2008, 42(7), 2323–2327.
7. T. D. Davies, et al., Prediction of episodic acidification in North-eastern USA: an empirical mechanistic approach, Hydrol. Processes, 1999, 13(8), 1181–1195.
8. A. Elshorbagy, R. S. V. Teegavarapu and L. Ormsbee, Total maximum daily load (TMDL) approach to surface water quality management: concepts, issues, and applications, Can. J. Civ. Eng., 2005, 32(2), 442–448.
9. 2000/60/Ec, D., Directive 2000/60/EC of the European Parlament and of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy, Official Journal of the European Communities, 2000, L327/1.
10. ANZECC, Australian Water Quality Guidelines for Fresh and Marine Waters, 2000, Australian and New Zealand Environment and Conservation Council, Canberra.
11. D. M. Robertson and E. D. Roerisch, Influence of various water quality sampling strategies on load estimates for small streams, Water Resour. Res., 1999, 35, 3747–3759.
12. R. D. Harmel, et al., Cumulative uncertainty in measured streamflow and water quality data for small watersheds, Trans. ASABE, 2006, 49(3), 689–701.
13. B. Müller, M. Reinhardt and R. Gachter, High temporal resolution monitoring of inorganic nitrogen load in drainage waters, J. Environ. Monit., 2003, 5(5), 808–812.
14. E. Bernhardt, et al., Advancing Biogeochemical Research in the Field Hydrological Sciences: The CUASHI Hydrological Measurement Facility - Biogeochemical Component (HMF/BGC), in CUASHI. 2006, p. 1–25.
15. C. J. Patton, Autonomous environmental water quality monitoring - The future of continuous flow analysis, Environ. Chem., 2006, 3(1), 1–2.
16. J. Buffle and G. Horvai, In Situ Monitoring of Aquatic Systems. Chemical Analysis and Speciation. IUPAC Series on Analytical and Physical Chemistry of Environmental Systems, Vol. 6, 2000, Wiley, Chichester, 623.
17. M. Varney., Chemical Sensors in Oceanography, 2000, Gordon and Breach, , Amsterdam, 333.
18. R. D. Down and J. H. Lehr, Environmental Instrumentation and Analysis Handbook, 2005, John Wiley & Sons, New Jersey.
19. K. S. Johnson, et al., Chemical Sensor Networks for the Aquatic Environment, Chem. Rev., 2007, 107(2), 623–640.
20. B. Müller, R. Stierli and R. Gachter, A low-tech, low-cost passive sampler for the long-term monitoring of phosphate loads in rivers and streams, J. Environ. Monit., 2008, 10(7), 817–820.
21. P. Jordan, et al., High-resolution phosphorus transfers at the catchment scale: the hidden importance of non-storm transfers, Hydrology and Earth System Sciences, 2005, 9(6), 685–691.
22. J. W. Kirchner, et al., The fine structure of water-quality dynamics: the (high-frequency) wave of the future, Hydrol. Processes, 2004, 18(7), 1353–1359.
23. B. A. Pellerin, et al., Assessing the sources and magnitude of diurnal nitrate variability in the San Joaquin River (California) with an in situ optical nitrate sensor and dual nitrate isotopes, Freshwater Biol., 2009, 54(2), 376–387.
24. P. J. Johnes, Uncertainties in annual riverine phosphorus load estimation: Impact of load estimation methodology, sampling frequency, baseflow index and catchment population density, J. Hydrol., 2007, 332(1–2), 241–258.
25. H. B. Glasgow and et al, Real-time remote monitoring of water quality: a review of current applications, and advancements in sensor, telemetry, and computing technologies, J. Exp. Mar. Biol. Ecol., 2004, 300(1–2), 409–448.
26. S. Rusjan, M. Brilly and M. Mikos, Flushing of nitrate from a forested watershed: An insight into hydrological nitrate mobilization mechanisms through seasonal high-frequency stream nitrate dynamics, J. Hydrol., 2008, 354(1–4), 187–202.
27. G. Harris and A. L. Heathwaite, Inadmissible evidence: knowledge and prediction in land and riverscapes, J. Hydrol., 2005, 304(1–4), 3–19.
28. S. E. Manahan., Environmental Chemistry, 7th edn 2000, Lewis, Boca Raton, London, New York, Washington DC, 875.
29. J. Bobacka, A. Ivaska and A. Lewenstam, Potentiometric ion sensors, Chem. Rev., 2008, 108(2), 329–351.
30. P. J. Worsfold, Challenges in the determination of nutrient species in natural waters, Microchim. Acta, 2006, 154(1–2), 45–48.
31. T. Le Goff, et al., A solvent-free method for preparing improved quality ion-selective electrode membranes, Green Chem., 2002, 4(5), 486–491.
32. T. Le Goff, et al., An accurate and stable nitrate-selective electrode for the in situ determination of nitrate in agricultural drainage waters, Analyst, 2002, 127(4), 507–511.
33. E. Pretsch, The new wave of ion-selective electrodes, TrAC, Trends Anal. Chem., 2007, 26(1), 46–51.
34. S. Gray, et al., Flow analysis techniques for spatial and temporal measurement of nutrients in aquatic systems, Environ. Chem., 2006, 3(1), 3–18.
35. K. S. Johnson and L. J. Coletti, In situ ultraviolet spectrophotometry for high resolution and long-term monitoring of nitrate, bromide and bisulfide in the ocean, Deep-Sea Res., Part I, 2002, 49(7), 1291–1305.
36. J. van den Broeke, G. Langergraber and A. Weingartner, Online and in situ UV/vis spectroscopy for multi-parameter measurements: a brief review, Spectroscopy Europe, 2006, 18(4), 15–18.
37. H. W. Jannasch, et al., The Land/Ocean Biogeochemical Observatory: A robust networked mooring system for continuously monitoring complex biogeochemical cycles in estuaries, Limnol. Oceanography Methods, 2008, 6, 263–276.
38. C. M. Sakamoto, K. S. Johnson and L. J. Coletti, Improved algorithm for the computation of nitrate concentrations in seawater using an in situ ultraviolet spectrophotometer, Limnol. Oceanography, 2009, 7, 132–143.
39. B. J. Privett, J. H. Shin and M. H. Schoenfisch, Electrochemical sensors, Anal. Chem., 2008, 80(12), 4499–4517.

---

Footnotes

† Part of a themed issue dealing with water and water related issues.

‡ Electronic supplementary information (ESI) available: Instrument selection: a case study. See DOI: 10.1039/b910156j

---