- Email: [email protected]
Integrated continuous water quality monitoring for the LOIS river syndromme
the science of the lbtaIEnvironment k.--t-l Le-‘.-Ih ELSEVIER
The Scienceof the Total Environment 194/l95 (1997)
1 1 I- 118
Integrated continuous water quality monitoring for the LOIS river programme J.G. Evans, P.D. Wass, P. Hodgson Institute
of Hydrology,
Crowmarsh
Gifford,
Wallindord,
Oxon
OX10
8BB,
UK
Abstract
This paper describes the installation and operation of an integrated system for the measurement of suspended sediments for the Land-Ocean Interaction Study (LOIS) Rivers Programme. The system is designed to log a continuous record of turbidity and stage and to provide for the acquisition of river samples automatically, according to programmable sampling strategies. These strategies allow for different sampling rates to be employed throughout an event, so that an appropriate and efficient match between the river conditions and the acquisition of samples can be made. The system components and operating procedures are detailed and results are presented that show the successful operation of the instrumentation. 0 1997 Elsevier Science Ireland Ltd. Keywords:
Suspended sediment; Turbidity; Automatic sampling; Continuous measurement
1. Introduction Of particular importance for the River-Atmosphere-coast Study (Rivers) (LOIS RACS(R)) Programme is the characterisation of the intermittent flood events that are most significant to the transport of sediments and chemical species into the coastal zone via rivers. In recent years automatic water quality measurement systems have been developed and implemented by several groups (e.g. Vivian and Quinton, 1993), with the aim of improving the quality of collected data. This paper briefly describes the combination of sensors, automatic sampler and data logging components into a new, integrated system, to provide
the quality of samples and continuous record required by the LOIS River Programme for suspended sediment monitoring. This system, known as the Wallingford Integrated System for Environmental monitoring in Rivers (WISER), can be adapted to include sensors for a wide range of water quality parameters and is therefore well suited to many other applications in water quality monitoring.
2. System components Each suspended sediment monitoring system installed for the LOIS RACS(R) Programme con-
0048-9697/97/$17.000 1997 ElsevierScienceIreland Ltd. All rights reserved. I’II
SOO48-9697(96)05387-9
J.G.
112
Evans
et al. 1 The Science
of the Total
Environment
194/l 95 (I 997)
111~ 118
(4
CRIO measurement and control module
Fig.
1. (a) Sediment
monitoring
system
schematic
sisted of a combination of a pressure transducer (for stage measurement), two turbidity sensors, automatic sampler(s) and a data logger. At some stations, a telemetry facility was also incorporated to enable the remote interrogation of
and
(b) sediment
monitoring
system
installation.
the system. Each of these components is described in more detail in the following paragraphs. Schematic diagrams of the integrated system and an installation are shown in Fig. l(a) and (b).
J.G.
Evans
et al. / The Science
of’ rhe Total
2.1. System integration, control and logging
The system was based around a CR10 measurement and control module (Campbell Scientific, Shepshed, UK). This provided analogue to digital conversion of sensor inputs, data processing and logging, and controlled outputs to trigger the sampler(s) and to turn the turbidity sensors on and off. The sensors were interrogated by the CR 10 every 10 s and data were logged every 15 min. Each logged value was the average of the last six readings. This was particularly useful where the pressure transducer was not located in a stilling well and was therefore subject to open channel surface wave effects. Data was downloaded either with a data storage module on-site or via a connection to a conventional telephone line and a modem, enabling full duplex communication and remote access to all CR10 functions.
Etxironment
194/l95
(1997)
1 I I-1
18
113
different sampling strategies to be used easily and meant that no sorting of samples was required. A further advantage was that samples were still obtained in the event of one sampler failing. The Epic 1011 is a compact, portable, 12V DC sampler which uses a vacuum pump system. Each unit was configured for multiple container sampling (24 bottles, 0.5 1 each). Samples for suspended sediment analysis were drawn through low-cost reinforced PVC hose and collected in plastic bottles. River water wash cycles were used to purge the hose before the sample was taken. Where chemical analysis was required, samples were collected through a Teflon-lined hose, with glass sample bottles. With the various demands of the different LOIS sites, the flexibility in sampler programming afforded by the Epic 1011 was essential. The sampling strategy, sampling frequency and trigger thresholds were controlled using the CR10 logger for specific rivers or applications.
2.2. Pressure transducer
A silicon strain gauge pressure transducer (Druck, Groby, Leicestershire, model PDCR 830), with a resolution of approximately 1 mm (depth of water), was used at each site to measure stage. 2.3. Turbidity sensors
Transmission-based and nephelometric (90” scatter) optical turbidity sensors were used at each site to achieve good performance during both high and low turbidity events. These were both manufactured by Partech Instruments, St. Austell, Cornwall (models IR40C and IR12LS). Each sensor incorporated a near infrared LED which was modulated at approximately 600 Hz to reduce effects due to variations in ambient light conditions. Wass et al. (1997) describe further details on the calibration and performance of the turbidity sensors. 2.4. Automatic water sampler
One or two Epic 1011 wastewater samplers (Montec, Manchester) were deployed at most sites. The use of two samplers at the same site not only increased sample capacity but also enabled
2.5. Water chemistry measurements
Hydrolab H20 (Hydrodata Systems, London Colney, UK) water quality monitors were installed at several sites in parallel with the sediment monitoring systems and used to log water temperature, pH, conductivity, dissolved oxygen and redox.
3. Installation
and maintenance
The sites under study and the equipment deployed are listed in Table 1. With a few exceptions, all installations for suspended sediment monitoring have been housed within buildings operated by the Environmental Agency (EA). Generally there were one or two samplers housed within the buildings, together with the CR10 control module and interface. The pressure transducer used to measure river stage was mounted in the stilling well of the gauging station where one was available, and securely in the river itself where this was not possible. Cabling was routed from enclosures to the river using flexible ducting where no suitable alternative existed.
114
J.G.
Evans
et al. / The Science
of’ the Total
Environment
194/195
(1997)
I II -118
Table 1 LOIS field equipment installations River
Turbidity
Upper Swale Lower Swale Ure Nidd Wharfe Ouse (Skelton) Derwent Aire Calder Don Trent Tweed (Boleside) Tweed (Norham) Tweed (Ormiston)
J J J J J J
and Stage
J J J J J J J
A variety of methods for mounting the turbidity sensors were employed, the relevant considerations being to obtain representative turbidity values, to comply with EA siting requirements, to allow for easy access for cleaning and maintenance and to protect against vandalism. To ensure the continuous and reliable operation of these systems, a full maintenance schedule was followed. Visits to each site were made every week during the warmer months to prevent the build up of biological matter (e.g. algae) on the turbidity sensor lenses. When samplers stood unused for long periods, regular checking was undertaken to ensure continued readiness for event sampling.
4. Results
To evaluate the performance of the Epic samplers, a comparison of the samples obtained using a United States Geological Survey (USGS) depth integrating manual sampler and those from the autosampler was made. In several tests, it was found that during successive sampling, waters from the Epics initially contained higher suspended sediment concentrations than the USGS reference. As the sampling sequence progressed, this over-representation rapidly became an underrepresentation of around 10% (Fig. 2). Initial over sampling is explained by the incomplete purging
Samplers
Telemetered
Hydrolab H20
J J J J J
J
J
J J J J J J
J J J J J J J
J J J J J J J J
of the inlet hose before sampling, a situation resulting in particulates deposited in the hose between samples being drawn into the water collected for analysis. After these particulates have been removed from the hose, the under-representation of suspended sediment may be expected where water is drawn through a hydraulic lift of several metres. In such cases, the velocity of the water in the pipe may be much reduced and can be less than the settling velocity of some particles, an effect generally more noticeable during conditions of high flow and high turbidity, when larger particles are in suspension and the density of the sample is higher. The period of deployment varied between rivers, but most equipment was installed by October 1994, and has been operational since. Although a variety of sampling strategies were tested, the strategy used at all LOIS RACS(R) sites, shown as a flowchart in Fig. 3, triggered a set of samples once the stage and nephelometric turbidity levels had both increased above threshold levels set for each site. This prevented the wasteful sampling of storm events low in suspended sediments. Sampling continued until a set was complete or either trigger value fell to below its threshold. The use of three sampling intervals for each event provided a method of discharge-related sampling, whilst also having an element of continuous fixed time interval sampling; both methodologies were required in the LOIS RACS(R) core programme.
J. G. Evans et al. i The Science of the Total Environment 194/195 (1997) 111-118
115
River Swale (Leckby) 40 piiiG$q
30
Note: the mean sediment concentration of USGS samples over this thirty minute period was 90.1 mg/l.
20
10
0
-10
-20 0
4
9
16
19
21
23
26
26
30
Minutes Elapsed Since First Sample Fig. 2. Comparison of Epic and USGS samplers.
Sampling intervals and trigger thresholds were optimised according to the season (or antecedent conditions) and the type of river under investigation. The River Swale exhibited a ‘flashy’ hydrological response, and therefore required a high sampling frequency, as short as 90 min between samples for short lived summer events. In contrast, a river such as the Trent at North Muskham, just downstream of Newark, responded more slowly and over a longer time
period, requiring sampling intervals of 240 min for a similar event. Data collected in the first year of deployment from such sampling regimes was encouraging. Fig. 4(a) and (b) show the flow, turbidity and suspended sediment concentration (determined by weighing the particulates collected by the automatic sampler) during a storm event on the River Swale at Leckby Grange and on the River Trent at North Muskham.
116
J.G.
Evans
9
et al. / The Science
of’ the Total
Environment
194/195
(1997)
111-118
Start program every 15 minutes
+
FALSE
Measure stage and tuftdii every ten .seconds, for one minute, then average the six readings
TRUE
FALSE
7
Increment
time t
Trigger automatic sampler
Y
FALSE
I
t
Increment n set t=o v
t
NOMENCLATURE t = time elapsed since the last sample was collected. n = number of samples collected since the start of a set of samples. Na, Nb, NC = user defined number of samples in each group. Na+Nb+Nc comprise a set of samples. Ta, Tb, Tc = user defined time intetWs between samples for each group of samples.
Fig.
3. Simplified
flow
diagram
of the measurement
5. Conclusions
The value of integrated automatic water quality systems arises through the acquisition of a continuous data record and related samples. This leads to improved quality of data, through having suit-
and control
program.
able reference points for the continuous record and to efficient use of available resources, when the collection of samples is triggered according to pre-determined river conditions. Future water quality monitoring programmes are likely to use such integrated systems. Appropriate refinements
J.G.
Evans
et al. / The Science
of’ the Total
Environment
1941195
(1997)
117
1 II -118
a. River Trent at North Muskham
c‘rn mg 2 Ii
300 200 100 0
1 -
’ [ ..I 26
I 28
I 30
1 32
I 34
I 36
I 38
36
38
0
b. River Swale at Leckby Grange
26
30
28
32
34
Julian Day (1995)
Ia! 0
suspended
Fig.
sedimentaxxentration
4. Sampled
events
- - - - flow
on rivers
Trent
and Swale.
-
turbidity
118
J.G.
Evans
et al. / The Science
of the Total
may include the use of other sensors (e.g. pH) to trigger the collection of samples and the implementation of more complex sampling strategies, such as according to rate of change of flow.
Emironment
194/195
(1997)
I I1 -118
systems described here and Mary Turner for her part in the electronics system concepts and design.
References Acknowledgements The authors wish to acknowledge Alan Warwick and his colleagues in the engineering workshop at the Institute of Hydrology, for their role in the mechanical design and installation of the
Vivian, B.J. and J.N. Quinton, 1993. Automated water sampling in ephemeral hydrological systems. Earth Surface Processes & Landforms, 18: 8633868. Wass, P.D., SD. Marks, J.W. Finch, G.J.L. Leeks and J.K. Ingram, 1997. River turbidity and remote-sensed imagery for suspended sediment transport in the Humber rivers, Sci. Total Environ., 194/195: 263-283.
The Scienceof the Total Environment 194/l95 (1997)
1 1 I- 118
Integrated continuous water quality monitoring for the LOIS river programme J.G. Evans, P.D. Wass, P. Hodgson Institute
of Hydrology,
Crowmarsh
Gifford,
Wallindord,
Oxon
OX10
8BB,
UK
Abstract
This paper describes the installation and operation of an integrated system for the measurement of suspended sediments for the Land-Ocean Interaction Study (LOIS) Rivers Programme. The system is designed to log a continuous record of turbidity and stage and to provide for the acquisition of river samples automatically, according to programmable sampling strategies. These strategies allow for different sampling rates to be employed throughout an event, so that an appropriate and efficient match between the river conditions and the acquisition of samples can be made. The system components and operating procedures are detailed and results are presented that show the successful operation of the instrumentation. 0 1997 Elsevier Science Ireland Ltd. Keywords:
Suspended sediment; Turbidity; Automatic sampling; Continuous measurement
1. Introduction Of particular importance for the River-Atmosphere-coast Study (Rivers) (LOIS RACS(R)) Programme is the characterisation of the intermittent flood events that are most significant to the transport of sediments and chemical species into the coastal zone via rivers. In recent years automatic water quality measurement systems have been developed and implemented by several groups (e.g. Vivian and Quinton, 1993), with the aim of improving the quality of collected data. This paper briefly describes the combination of sensors, automatic sampler and data logging components into a new, integrated system, to provide
the quality of samples and continuous record required by the LOIS River Programme for suspended sediment monitoring. This system, known as the Wallingford Integrated System for Environmental monitoring in Rivers (WISER), can be adapted to include sensors for a wide range of water quality parameters and is therefore well suited to many other applications in water quality monitoring.
2. System components Each suspended sediment monitoring system installed for the LOIS RACS(R) Programme con-
0048-9697/97/$17.000 1997 ElsevierScienceIreland Ltd. All rights reserved. I’II
SOO48-9697(96)05387-9
J.G.
112
Evans
et al. 1 The Science
of the Total
Environment
194/l 95 (I 997)
111~ 118
(4
CRIO measurement and control module
Fig.
1. (a) Sediment
monitoring
system
schematic
sisted of a combination of a pressure transducer (for stage measurement), two turbidity sensors, automatic sampler(s) and a data logger. At some stations, a telemetry facility was also incorporated to enable the remote interrogation of
and
(b) sediment
monitoring
system
installation.
the system. Each of these components is described in more detail in the following paragraphs. Schematic diagrams of the integrated system and an installation are shown in Fig. l(a) and (b).
J.G.
Evans
et al. / The Science
of’ rhe Total
2.1. System integration, control and logging
The system was based around a CR10 measurement and control module (Campbell Scientific, Shepshed, UK). This provided analogue to digital conversion of sensor inputs, data processing and logging, and controlled outputs to trigger the sampler(s) and to turn the turbidity sensors on and off. The sensors were interrogated by the CR 10 every 10 s and data were logged every 15 min. Each logged value was the average of the last six readings. This was particularly useful where the pressure transducer was not located in a stilling well and was therefore subject to open channel surface wave effects. Data was downloaded either with a data storage module on-site or via a connection to a conventional telephone line and a modem, enabling full duplex communication and remote access to all CR10 functions.
Etxironment
194/l95
(1997)
1 I I-1
18
113
different sampling strategies to be used easily and meant that no sorting of samples was required. A further advantage was that samples were still obtained in the event of one sampler failing. The Epic 1011 is a compact, portable, 12V DC sampler which uses a vacuum pump system. Each unit was configured for multiple container sampling (24 bottles, 0.5 1 each). Samples for suspended sediment analysis were drawn through low-cost reinforced PVC hose and collected in plastic bottles. River water wash cycles were used to purge the hose before the sample was taken. Where chemical analysis was required, samples were collected through a Teflon-lined hose, with glass sample bottles. With the various demands of the different LOIS sites, the flexibility in sampler programming afforded by the Epic 1011 was essential. The sampling strategy, sampling frequency and trigger thresholds were controlled using the CR10 logger for specific rivers or applications.
2.2. Pressure transducer
A silicon strain gauge pressure transducer (Druck, Groby, Leicestershire, model PDCR 830), with a resolution of approximately 1 mm (depth of water), was used at each site to measure stage. 2.3. Turbidity sensors
Transmission-based and nephelometric (90” scatter) optical turbidity sensors were used at each site to achieve good performance during both high and low turbidity events. These were both manufactured by Partech Instruments, St. Austell, Cornwall (models IR40C and IR12LS). Each sensor incorporated a near infrared LED which was modulated at approximately 600 Hz to reduce effects due to variations in ambient light conditions. Wass et al. (1997) describe further details on the calibration and performance of the turbidity sensors. 2.4. Automatic water sampler
One or two Epic 1011 wastewater samplers (Montec, Manchester) were deployed at most sites. The use of two samplers at the same site not only increased sample capacity but also enabled
2.5. Water chemistry measurements
Hydrolab H20 (Hydrodata Systems, London Colney, UK) water quality monitors were installed at several sites in parallel with the sediment monitoring systems and used to log water temperature, pH, conductivity, dissolved oxygen and redox.
3. Installation
and maintenance
The sites under study and the equipment deployed are listed in Table 1. With a few exceptions, all installations for suspended sediment monitoring have been housed within buildings operated by the Environmental Agency (EA). Generally there were one or two samplers housed within the buildings, together with the CR10 control module and interface. The pressure transducer used to measure river stage was mounted in the stilling well of the gauging station where one was available, and securely in the river itself where this was not possible. Cabling was routed from enclosures to the river using flexible ducting where no suitable alternative existed.
114
J.G.
Evans
et al. / The Science
of’ the Total
Environment
194/195
(1997)
I II -118
Table 1 LOIS field equipment installations River
Turbidity
Upper Swale Lower Swale Ure Nidd Wharfe Ouse (Skelton) Derwent Aire Calder Don Trent Tweed (Boleside) Tweed (Norham) Tweed (Ormiston)
J J J J J J
and Stage
J J J J J J J
A variety of methods for mounting the turbidity sensors were employed, the relevant considerations being to obtain representative turbidity values, to comply with EA siting requirements, to allow for easy access for cleaning and maintenance and to protect against vandalism. To ensure the continuous and reliable operation of these systems, a full maintenance schedule was followed. Visits to each site were made every week during the warmer months to prevent the build up of biological matter (e.g. algae) on the turbidity sensor lenses. When samplers stood unused for long periods, regular checking was undertaken to ensure continued readiness for event sampling.
4. Results
To evaluate the performance of the Epic samplers, a comparison of the samples obtained using a United States Geological Survey (USGS) depth integrating manual sampler and those from the autosampler was made. In several tests, it was found that during successive sampling, waters from the Epics initially contained higher suspended sediment concentrations than the USGS reference. As the sampling sequence progressed, this over-representation rapidly became an underrepresentation of around 10% (Fig. 2). Initial over sampling is explained by the incomplete purging
Samplers
Telemetered
Hydrolab H20
J J J J J
J
J
J J J J J J
J J J J J J J
J J J J J J J J
of the inlet hose before sampling, a situation resulting in particulates deposited in the hose between samples being drawn into the water collected for analysis. After these particulates have been removed from the hose, the under-representation of suspended sediment may be expected where water is drawn through a hydraulic lift of several metres. In such cases, the velocity of the water in the pipe may be much reduced and can be less than the settling velocity of some particles, an effect generally more noticeable during conditions of high flow and high turbidity, when larger particles are in suspension and the density of the sample is higher. The period of deployment varied between rivers, but most equipment was installed by October 1994, and has been operational since. Although a variety of sampling strategies were tested, the strategy used at all LOIS RACS(R) sites, shown as a flowchart in Fig. 3, triggered a set of samples once the stage and nephelometric turbidity levels had both increased above threshold levels set for each site. This prevented the wasteful sampling of storm events low in suspended sediments. Sampling continued until a set was complete or either trigger value fell to below its threshold. The use of three sampling intervals for each event provided a method of discharge-related sampling, whilst also having an element of continuous fixed time interval sampling; both methodologies were required in the LOIS RACS(R) core programme.
J. G. Evans et al. i The Science of the Total Environment 194/195 (1997) 111-118
115
River Swale (Leckby) 40 piiiG$q
30
Note: the mean sediment concentration of USGS samples over this thirty minute period was 90.1 mg/l.
20
10
0
-10
-20 0
4
9
16
19
21
23
26
26
30
Minutes Elapsed Since First Sample Fig. 2. Comparison of Epic and USGS samplers.
Sampling intervals and trigger thresholds were optimised according to the season (or antecedent conditions) and the type of river under investigation. The River Swale exhibited a ‘flashy’ hydrological response, and therefore required a high sampling frequency, as short as 90 min between samples for short lived summer events. In contrast, a river such as the Trent at North Muskham, just downstream of Newark, responded more slowly and over a longer time
period, requiring sampling intervals of 240 min for a similar event. Data collected in the first year of deployment from such sampling regimes was encouraging. Fig. 4(a) and (b) show the flow, turbidity and suspended sediment concentration (determined by weighing the particulates collected by the automatic sampler) during a storm event on the River Swale at Leckby Grange and on the River Trent at North Muskham.
116
J.G.
Evans
9
et al. / The Science
of’ the Total
Environment
194/195
(1997)
111-118
Start program every 15 minutes
+
FALSE
Measure stage and tuftdii every ten .seconds, for one minute, then average the six readings
TRUE
FALSE
7
Increment
time t
Trigger automatic sampler
Y
FALSE
I
t
Increment n set t=o v
t
NOMENCLATURE t = time elapsed since the last sample was collected. n = number of samples collected since the start of a set of samples. Na, Nb, NC = user defined number of samples in each group. Na+Nb+Nc comprise a set of samples. Ta, Tb, Tc = user defined time intetWs between samples for each group of samples.
Fig.
3. Simplified
flow
diagram
of the measurement
5. Conclusions
The value of integrated automatic water quality systems arises through the acquisition of a continuous data record and related samples. This leads to improved quality of data, through having suit-
and control
program.
able reference points for the continuous record and to efficient use of available resources, when the collection of samples is triggered according to pre-determined river conditions. Future water quality monitoring programmes are likely to use such integrated systems. Appropriate refinements
J.G.
Evans
et al. / The Science
of’ the Total
Environment
1941195
(1997)
117
1 II -118
a. River Trent at North Muskham
c‘rn mg 2 Ii
300 200 100 0
1 -
’ [ ..I 26
I 28
I 30
1 32
I 34
I 36
I 38
36
38
0
b. River Swale at Leckby Grange
26
30
28
32
34
Julian Day (1995)
Ia! 0
suspended
Fig.
sedimentaxxentration
4. Sampled
events
- - - - flow
on rivers
Trent
and Swale.
-
turbidity
118
J.G.
Evans
et al. / The Science
of the Total
may include the use of other sensors (e.g. pH) to trigger the collection of samples and the implementation of more complex sampling strategies, such as according to rate of change of flow.
Emironment
194/195
(1997)
I I1 -118
systems described here and Mary Turner for her part in the electronics system concepts and design.
References Acknowledgements The authors wish to acknowledge Alan Warwick and his colleagues in the engineering workshop at the Institute of Hydrology, for their role in the mechanical design and installation of the
Vivian, B.J. and J.N. Quinton, 1993. Automated water sampling in ephemeral hydrological systems. Earth Surface Processes & Landforms, 18: 8633868. Wass, P.D., SD. Marks, J.W. Finch, G.J.L. Leeks and J.K. Ingram, 1997. River turbidity and remote-sensed imagery for suspended sediment transport in the Humber rivers, Sci. Total Environ., 194/195: 263-283.