www.elsevier.com/locate/marpolbul

Marine Pollution Bulletin 51 (2005) 113–118

Rapid underway profiling of water quality in Queensland estuaries

Jonathan Hodge a,*, Ben Longstaff a, Andy Steven a, Phillip Thornton a,Peter Ellis b, Ian McKelvie b

a Water Sciences Unit, Environmental Protection Agency, Queensland, 80 Meiers Road, Indooroopilly, Brisbane 4068, Australiab Water Studies Centre, School of Chemistry, Monash University, Clayton 3168, Victoria, Australia

Abstract

We present an overview of a portable underway water quality monitoring system (RUM-Rapid Underway Monitoring), devel-

oped by integrating several off-the-shelf water quality instruments to provide rapid, comprehensive, and spatially referenced �snap-shots� of water quality conditions. We demonstrate the utility of the system from studies in the Northern Great Barrier Reef

(Daintree River) and the Moreton Bay region. The Brisbane dataset highlights RUM�s utility in characterising plumes as well asits ability to identify the smaller scale structure of large areas. RUM is shown to be particularly useful when measuring indicators

with large small-scale variability such as turbidity and chlorophyll-a. Additionally, the Daintree dataset shows the ability to inte-

grate other technologies, resulting in a more comprehensive analysis, whilst sampling offshore highlights some of the analytical

issues required for sampling low concentration data. RUM is a low cost, highly flexible solution that can be modified for use in

any water type, on most vessels and is only limited by the available monitoring technologies.

� 2004 Elsevier Ltd. All rights reserved.

Keywords: Water quality; Technology; Estuaries; Australia

1. Introduction

Effective monitoring and reporting against environ-

mental objectives is key to evaluating outcomes of

catchment and coastal action plans. In Queensland,

plans to mitigate the risk of catchment-derived inputs

of nutrients, sediment and toxicants detrimentally affect-

ing the integrity of the iconic Great Barrier Reef andMoreton Bay ecosystems are articulated in the Reef

Water Quality Protection Plan (RWQPP, 2003) and

the South East Queensland Regional Water Quality

Management Strategy (MBWCP, 2001). As well as set-

ting a blueprint for implementing a broad mix of catch-

ment-based actions—from regulatory and planning

frameworks to self-management, economic incentives

and extension—these initiatives establish a range of

0025-326X/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.marpolbul.2004.10.043

* Corresponding author. Tel.: +61 7 389 69240.

E-mail address: jonathan.hodge@epa.qld.gov.au (J. Hodge).

short- to long-term management and ecosystem-based

objectives. Such plans require comprehensive monitor-

ing of water quality parameters to inform on the success

or otherwise of management intervention activities.

However, there is growing unease amongst scientists,

natural resource managers and politicians about the lim-

itations of traditional manual monitoring of water qual-

ity at fixed sites. Site-based monitoring is costly andoften fails to resolve the dynamic spatial complexity of

coastal waters, leaving a high degree of uncertainty as

to whether management objectives are being met, or

what the full extent of significant events such as floods

or spills are on coastal systems.

A variety of technologies are now available for the

rapid acquisition of water quality data that potentially

provide the means for cost-effective, comprehensivemonitoring: sondes and CTDs for measuring tempera-

ture, salinity, pH and dissolved oxygen; fluorometric

technologies for chlorophyll biomass and phytoplankton

114 J. Hodge et al. / Marine Pollution Bulletin 51 (2005) 113–118

composition; flow injection and loop-flow analysis for

the acquisition of some nutrient species; and acoustic

doppler-based devices for current profiling.

Adoption of these technologies for routine monitor-

ing has been relatively slow in Australia because of pre-

vailing perceptions that they are costly, require largevessels for deployment, have poor analytical resolution,

are difficult to maintain and require detailed technical

capability. A further challenge is that there are only a

few off-the-shelf technologies that cost-effectively inte-

grate these various components.

Here, we present an overview of a portable underway

water quality monitoring system (RUM: Rapid Under-

way Monitoring) developed by integrating several off-the-shelf water quality instruments to provide rapid,

comprehensive, and spatially referenced �snapshots� ofwater quality conditions. We demonstrate the utility of

the system from studies in the Daintree River (North

Queensland) and the Brisbane River (Moreton Bay,

Queensland).

2. Methods

2.1. Instrumentation

RUM is comprised of a number of core components

with the flexibility to add on other instruments yet pro-

Fig. 1. Schematic diagram of the Rapid Underway Monitoring system illus

pump used to deliver water to the onboard instrumentation, (b–d) suite of onb

side of the vessel, (g) GPS unit and computer components.

duce a single integrated data output, whilst presenting

this visually in near real-time (Fig. 1).

Customised LabVIEWTM instrument control software

communicates with instruments, controls data flow,

visualises and stores data. A GPS is used to spatially ref-

erence the incoming data strings and the combined dataare presented on the computer screen in a variety of user

selected graphical and tabular formats.

Communication with scientific instruments is via

serial (RS-232 or RS-485) connections, and multiple

4-port serial to USB hubs serve to increase the number

of serial connections possible on a standard laptop.

These devices allocate static port numbers, ensuring

consistency when the configuration of instruments andhubs is changed. Data acquisition speed can be limited

by the GPS unit being used; our GPS delivered data

every 2s but it is possible to get GPS units which acquire

data at higher rates.

The core system utilises a YSITM multi-parameter

probe (YSI 6600) connected to a YSITM flow cell to mea-

sure basic physico-chemical parameters: pH, tempera-

ture, specific conductance, dissolved oxygen, turbidityand chlorophyll-a concentration as fluorescence. Other

instruments can be easily added depending on the pur-

pose of the study being conducted. Routinely, RUM

also uses a BBETM fluoroprobe (for measuring concentra-

tions of different algal groups), an underwater video

camera and an RDITM acoustic doppler current profiler

trating key components and the required integration. (a) Diaphragm

oard instruments currently trialled, (e–f) instruments mounted over the

J. Hodge et al. / Marine Pollution Bulletin 51 (2005) 113–118 115

(ADCP, Workhorse Sentinel model). The developmen-

tal flow injection nutrient analyser (Ellis et al., 2003;

Lyddy-Meaney et al., 2002) has also been trialled. Each

component of the system is modular in design, from the

USB hubs to the structure of the programming within

LabVIEWTM. Each instrument and its related LabVIEWTM

module can be added, removed, duplicated or modified

depending on which instruments are connected to the

laptop.

Whilst underway, sub-surface water is pumped to the

instrumentation using a diaphragm pump. During sam-

pling the intake was positioned approximately 1m

below the surface to avoid turbulence created by boat

wash and to ensure that it remained submerged at alltimes. Reinforced tubing was used to ensure pressure

does not collapse the tubing. The system was designed

so that any bubbles entering the flow cells could easily

be removed. The flow cell was shaded to limit interfer-

ence of ambient light on the optical turbidity and fluo-

rescence probes.

2.2. Field trials

Underway monitoring was conducted in estuarine

areas of the Brisbane River in Southeast Queensland

and the Daintree River in the Northern Great Barrier

Reef (Fig. 2).

At both locations, �dry� and �wet� season sampling

was conducted in a continuous transect upstream from

the river mouth. The Brisbane River dry season sam-pling in December 2003 extended �40km upstream

(Fig. 3a), and in January 2004, following a wet event,

�70km upstream (Fig. 3b). In the Daintree River, dry

season sampling in October 2003 was limited to �5km

Fig. 2. Extent of sampling in Daintree (a) and Brisbane Rivers (b).

Dry and wet sampling runs were conducted in both systems.

from the river mouth (Fig. 4a), but extended �9km up-

stream during wet sampling in April 2004 (Fig. 4b).

Sampling at this time was not continuous with a gap

in data collection from �5 to 5.5km representing a dif-

ference of �2h. As water quality did not appear to

change significantly during this period, data are pre-sented on a single graph.

In the Brisbane River, spatial interpolations of tur-

bidity were conducted on the Brisbane River data.

Raw data was extracted from the underway results

at regular monitoring locations and interpolated val-

ues were calculated between these points using locally

weighted regressions (SPLUS 6, 2001; Cressie, 1993).

Annual average flow is �0.9km3 in the BrisbaneRiver and �0.95km3 in the Daintree River. Both rivers

have a number of tributaries, and the Brisbane River has

eight wastewater discharges treating approximately 1.25

million equivalent persons. The Daintree River has no

major point source discharges.

3. Results

3.1. Brisbane River

During dry weather sampling, salinity and turbidity

showed strong linear gradients with relatively little

small-scale variability, salinity being affected by local-

ised inputs from minor tributaries and turbidity from

these inputs as well as tidal mixing of fine-bed sediments(Fig. 3a). Near the river mouth, a sharp localised drop in

salinity and an increase in both turbidity and chloro-

phyll-a occurred adjacent to a major wastewater treat-

ment plant (WWTP) discharge.

During the wet season sampling—conducted during a

rain event—salinity was �2 PSU lower than during the

dry season at comparable distances from the river

mouth, with significant drops from 5 to 10km (Fig.3b). Compared to the dry season, there was significant

smaller scale structure in both the salinity and turbidity

data and the broad-scale structure was less discernible,

particularly for turbidity. Significant peaks in turbidity

corresponded with reductions in salinity and occurred

opposite confluences with tributary creeks, such as Bul-

imba (�7km), Oxley (�40km) and Wolston Creeks

(�60km).

3.2. Daintree River

Dry period monitoring revealed little change in water

quality from marine conditions in the lower 5km of the

estuary, though some mixing with fresher waters oc-

curred above 4km from the mouth (Fig. 4a).

Wet season concentrations of salinity and turbidity(measured �2 weeks after significant flows) were higherand more variable than during the dry season (Fig. 4b).

Fig. 3. Turbidity and salinity data for dry (a) and wet (b) period sampling (December 2003 and January 2004, respectively) in the Brisbane River.

Minimal variability was seen in both indicators during the dry with sharp spikes in turbidity and salinity near the mouth. High small-scale variability

was evident during the wet, related to inputs from tributaries including Bulimba (7km), Oxley (40km) and Wolston Creeks (60km).

116 J. Hodge et al. / Marine Pollution Bulletin 51 (2005) 113–118

Turbidity levels were considerably higher in the lower

5km of the estuary than during the dry season. Turbid-

ity maxima (>80 NTU) occurred near the mouth, sug-

gesting that the dominant source of sediment in the

Daintree River during this period was wind- and tide-

driven resuspension from waters outside the mouth of

the estuary. A localised reduction of both turbidityand salinity occurred where the southern arm of the

Daintree joins the main channel, 4–5km upstream of

the mouth; above this point there was a more rapid tran-

sition to fresher, clearer water. Underway analysis of

soluble reactive phosphorus (SRP) concentrations indi-

cate minor variability throughout the system with a con-

servative mixing pattern along the estuary, indicating

little uptake of phosphorus along the estuary (Fig. 4b).

4. Discussion

By combining off-the-shelf instrumentation and soft-

ware, a cost-effective and readily deployable system that

rapidly measures water quality has been developed. The

basic configuration of RUM, including software, GPS,sonde, computer hardware and computer software can

be assembled for �AUD $30 000. However, one of the

key features of RUM is the ease with which additional

instrumentation can be added. We routinely deploy an

ADCP to provide hydrodynamic information and a

BBETM fluorometer to indicate the relative taxonomic

composition of phytoplankton. We have deployed

RUM from boats as smalls as 4m to vessels greater than

15m with relative ease, and acquired data at water

speeds of greater than 25 knots (depending on which

instruments are deployed). A key consideration at these

high speeds, or during rough conditions, is to ensureundue turbulence does not result.

A GPS unit that can output data to a serial connec-

tion is crucial to ensure that data can be referenced to

common spatial coordinates. GPS units vary in their

ability to output data and the overall frequency at which

data can be acquired is sometimes limited by the rate at

which the GPS can update. The analytical detection lim-

its of the equipment are also a consideration, particu-larly in reef waters where turbidity and chlorophyll-a

concentrations were often below the analytical limits

of our equipment. This is easily remedied by using better

resolution probes, without necessarily having to upgrade

the equipment. However, the potential applications for

RUM will generally be in waters of relatively high con-

centrations. These include monitoring of flood or mixing

zone plumes, or optimising and validating monitoringstrategies through identifying the major sources of spa-

tial and temporal variation.

One specific application demonstrated in our field tri-

als is the potential for assessing mixing zone size and

compliance. RUM passed through the discharge plume

Fig. 4. Results for dry (a) and wet (b) period sampling (October 2003 and April 2004, respectively) in the Daintree River. Dry period sampling shows

near-oceanic conditions extending almost 5km upstream. Wet period shows more variable results with a greater freshwater influence. Turbidity

maxima occur near the mouth, indicating that the turbid water is related to wind and tide driven resuspension of sediment. Conservative mixing of

phosphate (SRP) indicates that little uptake is occurring along the estuary.

Fig. 5. Mouth of the Brisbane River during dry period sampling

showing an increase in turbidity, an increase in chlorophyll-a and

decrease in salinity adjacent to a major wastewater treatment plant

discharge.

J. Hodge et al. / Marine Pollution Bulletin 51 (2005) 113–118 117

of Brisbane�s largest wastewater treatment plant, Lug-gage Point, and accurately captured the impact of the

discharge on the ambient environment (Fig. 5). At the

point of discharge there were spikes in chlorophyll-a

and turbidity and a dip in salinity. The increased levels

returned to ambient conditions within a few hundredmeters of the discharge pipe. This highlights the ability

of RUM to identify fine-scale spatial patterns in water

quality. In order to fully characterise the dispersion

characteristics of the plume, additional transects could

easily be taken and processed in a GIS package.

A further use demonstrated by the Brisbane data is

the ability to detect small-scale structure that might

otherwise be �missed� by discrete sample site methodolo-gies. Fig. 6 compares the spatial structure of turbidity

measured by RUM under wet and dry conditions with

spatial interpolations of turbidity made from discrete

monthly sampling at 12 sites in the Brisbane River as

part of the Ecosystem Health Monitoring Program

(EHMP, 2004).

With the exception of the region around the WWTP

discharge it is evident that during the dry season, whenvariation is minimal, routine monitoring of sites cap-

tures the spatial structure of water quality and can pro-

vide robust assessments of water quality condition (Fig.

6a). In comparison, site-based monitoring under wet

conditions potentially misses much of the spatial vari-

ability measured using RUM. In these types of condi-

tions, spatial prediction may fail to provide a reliable

interpretation of the raw data, as the sampling sites fre-

quently do not correspond to peaks in the data and thus

Fig. 6. Turbidity results from the Brisbane River showing data extracted at routine monitoring sites and spatial predictions based on these data

points. Dry period results match site-based monitoring and the prediction well. Site-based monitoring and predictions underestimate underlying

conditions during the wet period.

118 J. Hodge et al. / Marine Pollution Bulletin 51 (2005) 113–118

the impact of wet events may be underestimated. Fig. 6b

shows that neither of the two sharp spikes in turbidity(49 and 60km) would have been identified and the

broader spikes at around 30 and 40km were not cap-

tured effectively. Underway monitoring, however, pro-

vides a detailed view of the small-scale structure of the

data, where problem areas originate and the spatial ex-

tent of the impact of individual inputs.

Additional applications of RUM could include

deployment on commercial vessels such as ferries(Swertz et al., 1999; Holley and Hydes, 2002), investiga-

tions of environmental incidents such as oil spills, and

sewage overflows and baseline monitoring of previously

unmonitored systems.

Future directions for the enhancement of RUM in-

clude the use of mini computers with wireless LCD

touch-screen monitors, modification of waterproof cases

to minimise cabling issues and the use of peristalticpumps in place of the existing diaphragm pump.

Acknowledgment

The following people assisted us in the development

of this system: John Ferris, Ray Clark, Paul Maxwell,

Jeff Shortell, Peter Toscas, Todd Nelson and DarrenHanis. We thank them for their advice and support.

References

Cressie, N.A.C., 1993. Statistics for Spatial Data, revised edition. John

Wiley and Sons, New York.

EHMP, 2004. Ecosystem Health Monitoring Program 2002–2003

Annual Technical Report. Moreton Bay Waterways and Catch-

ments Partnership, Brisbane.

Ellis, P.S., Lyddy-Meaney, A.J., Worsfold, P.J., McKelvie, I.D., 2003.

Multi-reflection photometric flow cell for use in flow injection

analysis of estuarine waters. Analytica Chimica Acta 499, 81–

89.

Holley, S.E., Hydes, D.J., 2002. �Ferry-Boxes� and data stations for

improved monitoring of eutrophication-related processes: applica-

tion in Southampton Water UK, a temperate latitude hypernutri-

fied estuary. Hydrobiologia 475/476, 99–110.

Lyddy-Meaney, A.J., Ellis, P.S., Worsfold, P.J., Butler, E.C.V.,

McKelvie, I.D., 2002. A compact flow injection analysis system

for surface mapping of phosphate in marine waters. Talanta 58,

1043–1053.

MBWCP, 2001. South-east Queensland Regional Water Quality

Management Strategy. Moreton Bay Waterways and Catchments

Partnership, Brisbane, Australia.

RWQPP, 2003. Reef Water Quality Protection Plan. Queensland

Government, Brisbane, Australia.

SPLUS 6, 2001. SPLUS 6 for Windows Guide to Statistics. Vol. 1.

Insightful Corporation, Seattle, WA.

Swertz, O.C., Colijn, F., Hofstraat, H.W., Althuis, B.A., 1999.

Temperature, salinity and fluorescence in southern North Sea:

high-resolution data sampled from a ferry. Environmental Man-

agement 23 (4), 527–538.