The field of water analysis is entering a new area, with much talk of data sharing, new testing devices and water quality forecasting, as Emma Davies discovers

The field of water analysis is entering a new area, with much talk of data sharing, new testing devices and water quality forecasting, as Emma Davies discovers



Lake 658 is in the Experimental Lakes Area in Ontario, Canada

Lake 658 in Ontario, Canada, bears all the hallmarks of a dream destination for the keen fisherman - deep blue water full of fish, complete peace, and forests as far as the eye can see. But only a foolhardy fisherman would eat his catch - it is highly likely to be heavily contaminated with mercury. 658 is in fact an experimental lake to which mercury has been deliberately added so that its movement can be monitored. It’s not the only idyllic location whose waters are the subject of intense scientific scrutiny - teams of scientists the world over are working on water analysis in remote locations. Often you’ll have to look hard to find any evidence of the scientists - many are developing remotely operated analytical systems so that an antenna poking out of the water is all that’s visible.

The Lake 658 mercury project has a name that sounds like a heavy metal band and a logo to match - Metaalicus, which stands for Mercury experiment to assess atmospheric loading in Canada and the US. ’We’re attempting to demonstrate whether or not ecosystems respond to a change in mercury loading,’ says David Krabbenhoft, a geochemist at the US Geological Survey in Wisconsin, who works on the project. Between 2001 and 2007, Metaalicus researchers loaded the lake and its drainage basin with traceable and stable non-radioactive isotopes of inorganic mercury (Hg(ii)). They added different isotopes to the upland, wetland and lake surface (200Hg, 198Hg and 202Hg respectively) to determine the relative contributions of these sources to fish mercury levels. ’We look at mercury levels in the water and in the entire food web - all the way from the single cell algae at the bottom to the large fish at the top,’ says Krabbenhoft. A team of scientists is at the lake site all year round monitoring the mercury levels.

If you were to drink the water in the lake it would not be toxic - the mercury levels are very low. ’But there’s this really startling thing that happens with mercury - it biomagnifies,’ explains Krabbenhoft. Mercury has the largest biomagnification factor of any contaminant, with concentrations in fish up to 10 million times higher than mercury levels in the water. 

Mercury affects the function of neurological systems by attaching to proteins. It can also harm the reproductive cycles of fish and birds. ’It’s a bit harder to prove with humans because studies are confounded by the fact that humans mostly don’t eat fish that are caught locally,’ says Krabbenhoft. He is particularly concerned about mercury exposure in foetuses. ’The problem is that the concentration of mercury in umbilical blood is typically about six times higher than that in maternal blood. When you divide the levels that are generally regarded as safe by six you can’t even find fish with mercury levels that low.’ Krabbenhoft strongly believes that, if possible, pregnant women should not eat any fish. 

This is the first year that the team has not added any mercury to the lake. ’It may be the most important summer of all for the project,’ says Krabbenhoft. ’We fully expect to see a marked decline in the appearance of the isotopes in the water. Most of us working on the project feel that mercury doesn’t cycle for that long in these systems but that it gets locked up in sediments with very stable bonds to sulfur.’ But will the mercury come out of the sediment or stay in it? ’There’ll be a little of both,’ says Krabbenhoft. ’We’ll know 90 per cent of the answer by November 2008.’ 

The research team uses a mass spectrometer (MS) for most of its analysis. ’We customise the MS for mercury work by adding a purge and trap system,’ reveals Krabbenhoft. ’We trap the mercury onto a column that has gold beads in it. When the column is heated, the mercury comes off very suddenly, giving a loud signal.’ The team also uses atomic fluorescence to analyse bulk mercury. ’It’s a simple little atomic fluorescence box that you can buy for about US$2000 (?1070) but it is a far more sensitive piece of kit than the US$250 000 MS,’ says Krabbenhoft.

Sit back and wait 

The downside of needing an MS is that the researchers have to carry samples back to the lab for analysis. The prospect of doing full on-site analysis is very attractive to environmental scientists. What’s even more appealing is remote analysis, where data are collected using automated equipment and results are transmitted straight to the lab. 

Paige Novak, from the department of civil engineering at the University of Minnesota, US, and her colleagues, have set up such a system to study pollutants in storm water using commercially available wireless technologies, probes and sensors. 

Increasing urbanisation in the Minnesota region is damaging the water quality of streams that flow into lakes and increasing the concentrations of nutrients and organic chemicals, says Novak. Deciding how to deal with the urban streams is going to require plenty of data, she adds. Since last summer, her team has tested its water quality monitoring network in various test bed sites on the Upper Mississippi river, partly to see how well it can predict when levels of chemical and biological contaminants in urban streams will increase.  

The experimental part of the system is solar-powered and has three components. Every minute a Hydrolab sonde measures water temperature, pH, turbidity and dissolved oxygen concentration. Meanwhile, a MicroLAB nutrient analyser measures nitrate and phosphate levels at 20 minute intervals, to help the researchers assess how concentrations match those of urban and agricultural pesticides in the water. Hydrolab and MicroLAB results are transmitted to the lab through a mobile telephone modem and then uploaded onto a website. 

The system also has a sample grabber, which can be operated remotely. Novak and her team currently take samples to monitor levels of the herbicide prometon and of caffeine, which highlights when water has been contaminated by urban sources. 

The Minnesota team is working with computer scientists to develop tools to detect when a potentially ’interesting’ event, such as a flood or water contamination, has occurred. ’This is a challenging problem because you’re comparing two or more data streams,’ says Novak. ’It would be really helpful if we can develop some predictive correlations.’  


A USGS researcher prepares sediment samples on a small boat on Lake 658

Share and share alike 

Novak says that a change in attitude will be needed among water researchers, so that data are collected openly and continuously. ’We want to focus on understanding fundamental processes and to ask open-ended questions,’ says Novak. ’This "big picture" approach is a relatively new way of looking at things in water analysis,’ she adds. ’The idea is to elevate water quality prediction to something similar to weather prediction.’ 

Other researchers share this vision and have set up the Waters Network - a joint initiative which spans several different research disciplines, funded by the National Science Foundation. Bill Arnold is part of the Minnesota team and has been and is involved in the Waters Network. ’The ultimate goal is to continuously predict water quality and quantity everywhere [in the US],’ he says. The network hopes to find better ways to collect and share data to help solve water problems, for example reducing the impact of floods and water shortages. ’The hope is that having a network will help us to create models and improve our predictive capabilities,’ says Arnold.  

The network currently has 11 test bed sites across the US, which it uses to help develop ways of monitoring and predicting water quality. By 2016, the Waters Network hopes to operate a group of fully functional observational and experimental facilities. 

Data sharing between different research groups will play a key role in the Waters Network. However, scientists are not always willing to share data, especially in areas such as specialised sensor development, warns Arnold. ’We may need to have certain rules, such as giving a period of exclusivity to publish for someone who collects the original data,’ he adds. ’Some kind of new paradigm has to come out of this.’ 

Europe does not yet have a water network. But, in the UK, a virtual observatory was approved for funding in July by the Natural Environment Research Council (Nerc), and will be up and running by the Autumn. It will be run by Louise Heathwaite, co-director of the Centre for Sustainable Water Management at Lancaster Environment Centre. The observatory will act as a platform for sharing data and methodologies, and for collectively interpreting results. 

Heathwaite hopes eventually to build a collaboration with the US Waters Network. ’We will set up a pilot in the first instance, which is likely to include a number of riverbasins,’ she says. ’The aim is to link together observation-data synthesis and modelling communities, to ensure wide access to available data.’ 



A researcher checks the grab sampler in an experimental pond in Minnesota

In the field

So it would seem that change is afoot in the field of water analysis. It all sounds exciting but many researchers remain frustrated at the lack of inexpensive portable water analysis devices. ’A lot of devices just don’t work well in the natural environment and the ones that do only measure routine compounds,’ says Novak. Few are really handheld and they are also pretty expensive, costing anything from $5000 - $15 000 apiece, she adds. What’s more, it takes a lot of time to keep the devices running well.  

Upal Ghosh, from the department of civil and environmental engineering at the University of Maryland, US, agrees. He is working, with his colleagues from chemical engineering, to change this situation, and has developed a small prototype device for testing pH, turbidity and oxygen concentration remotely.  

’Oxygen sensing is generally based on electrodes, but the problem is that they’re bulky and get fouled up. You have to clean and recalibrate them. There are some sensors that are less prone to fouling but they’re very expensive,’ says Ghosh. To help address this problem, he and his team have developed a small prototype sensor platform (about 6cm square). The oxygen sensor works by shining a particular wavelength of light produced by a light emitting diode onto a ruthenium dye immobilised on a silica gel. Exposing the dye to oxygen causes a phase shift in the light emitted which then triggers the sensor.  

The sensors transmit data through an antenna poking out of the water to a base station up to one mile away. In future the device could also include sensing elements to detect contaminants such as heavy metals or toxic gases, says Ghosh. He anticipates that it will take two years to commercialise his device. ’Right now we’re writing funding proposals so that we can test the prototypes over a long period of time,’ he says. He envisages that large numbers of his sensor platforms could be used over a wide monitoring area to gauge the sources and dynamics of pollution in large water bodies. 

Ghosh could perhaps seek advice from Bill Davison, from the aquatic chemistry research group at Lancaster University, UK, who has commercialised a simple water testing device, which now sells worldwide for just ?10 apiece. He and fellow researcher Hao Zhang developed the DGT (diffusive gradients in thin films) device in 1993. DGT accumulates dissolved substances in a controlled way and is routinely used to measure 10 metals, including iron, mercury, lead, copper, and zinc. It contains a layer of a binding agent impregnated in a hydrogel. On top of this is a diffusive hydrogel layer and a filter. Metal ions diffuse through the filter and the hydrogel to reach the resin layer. Different binding agents can be used for different metals and there is no need to calibrate. Once back in the lab, the binding layer is removed, eluted with a dilute acid, and the metal ions measured using a technique such as ICP-MS (inductively coupled plasma mass spectrometry).  

In 1997, Davison and Zhang set up a company called DGTResearch to sell the throwaway DGT devices. The company isn’t run in the strictest commercial sense, says Davison. ’We are more interested in supplying DGT to other labs around the world.’ Davison and Zhang have always shared DGT data with other researchers. ’We could have billed DGT as a "magic device" but from the outset we decided to publish freely on DGT’s positive and negative aspects,’ he recalls. Many other labs around the world have tested and used DGT and there now exists a ’collective knowledge,’ says Davison. ’Keeping everything open is vital and better for the company in the long run.’ 

Many water researchers are keen to see their devices and knowledge put to use in the developing world. ’We really need to start thinking about better water analysis and management and how it would work in the developing world,’ says Novak. ’If we can get better at handling things we will be able to deal with problems proactively.’ With climate change causing increasing numbers of floods and contamination episodes around the globe, the idea of being able to predict when crises are going to hit is certainly very attractive. 

Emma Davies is a freelance science writer based in Bishop’s Stortford, UK