HIGH FLYERS THINK TANK

Innovative technical solutions for water management in Australia

University of Adelaide, 30 October 2006

Information technology sciences
by Dr Alex Zelinsky, CSIRO ICT Centre, Canberra

In talking about another aspect of the water situation – the use of technology – I want to begin by making a couple of comments about something Max Brennan referred to: the multidisciplinary aspects of the whole water situation and challenge. In CSIRO we have a dedicated set of professionals who look at land and water issues, in the Division of Land and Water, but now we also have information and communication technology specialists, mathematicians, statisticians, industrial physicists looking at sensor network technologies. In fact, we are also looking at the atmospheric research. We are really looking at this whole water system and associated challenges in a holistic way.

The CSIRO National Research Flagship program is really about getting the 6500 people working within the organisation to collaborate on multi-disciplinary projects, and with the Water for a Healthy Country Flagship we have about $50M pa of the resources of the organisation starting to be pulled into this very significant area. The goal for the Flagship is to answer the question, 'What can we do by 2020?' There is a big stretch goal there: can we obtain a tenfold improvement in the utilisation of water, from an environmental point of view, a social point of view and also an economic point of view? So, this is a great challenge.

All the projects that are funded within the construct of this Flagship must contribute to this goal. Everything is outcome based, at CSIRO we are now all pushing towards this type of work. Today I am just talking about one aspect of that big program, the information technology aspect.

(Click on image for a larger version)

This slide introduces and summarises the water challenge. I won't talk much about this; I think everyone knows what are the key issues that we face.

(Click on image for a larger version)

One of the things we do know is that in meeting the challenge we need new policies and we also need new processes. To implement new processes and achieve better management outcomes of the resources, you actually have to understand what resources you have got. You can't manage, control or properly allocate water unless you know how much water you have got. This means we must be able to measure how much water we have. The big project we are involved in at CSIRO is the Water Resources Observation Network (WRON), which is about building new infrastructure that allow us to understand how much water we have got, allowing us feed forward this number into the whole system.

(Click on image for a larger version)

We don't actually have good snapshots of the current state of our water resources. For example consider the Murray-Darling Basin authority, we have been told reliably, that they believe the error in their measurements is about 1.1 gigalitres! That figure is only in one area, so how much water is there really in the whole Australian water system?

If we had a good snapshot and we also had forecasting models about water consumption, along with knowing what is happening with the climate – our climate models are becoming more sophisticated – we can start to predict how much water will be available in the future. Therefore, for water that could be allocated in the future, this means you can potentially put an economic model around the issue, you could trade the water, and you can then ensure efficient utilisation in a proper economic framework.

(Click on image for a larger version)

At CSIRO we are very much interested in meeting these challenges from an information perspective. This is just one part of the program; there are many other parts.

Interestingly, there does not exist a single national account of how much water we have. This information is basically distributed across state boundaries and also across the authorities and agencies. There are about 180-odd agencies each one of them has its own piece of information.

And there are no standards. A lot of the data sets are historical going back many years; people have logged the data – shared it in fact, but not in a common way. There is no easy way to aggregate the data, nor are there methods to store the data in a centralised place or make it available to others.

There are also problems with large gaps in the data sets, and a lot of the data may have been recorded on very sparse temporal basis: once a day sometimes once a week. What we are talking about here is really to have a particular sense of the value of data in real time.

Also, there are no automatic reporting or forecasting systems in place. There are mainly manual systems, which are very time consuming and expensive to produce.

So we have come at this problem from an information and communications technology perspective, asking what could be done with this problem if we change the game by being able to provide information in real time, at the right time, for the right people, and allow scientists and users of the water resources to actually understand what is the state of the system and then be able to use that information accordingly.

(Click on image for a larger version)

Basically, it is a water puzzle. We can look at various aspects of the puzzle. We do have ideas of how much water our irrigators use, we have an idea about what goes on in our climate, we do know about contaminants and we know about other aspects of urban water usage et cetera. But this is really is just pieces of a larger jigsaw puzzle.

(Click on images for larger versions)

What we are talking about it as the whole environment. It is not just different systems; they are all connected. And what we are very interested in doing is actually connecting up all the pieces – in other words, getting information from the various sources and linking them up –using modern telecommunications systems such as satellites, mobile comms, to do this. In a country the size of Australia, you're not going to actually link these parts of the jigsaw puzzle by say laying new optical fibre. However, wireless communications and satellite communications systems are becoming cheaper. There are new developments in these technology spaces in which we are very interested. We believe that you can actually deploy a lot of these systems very cheaply and then link them up through wireless communications.

So there is the infrastructure. We are interested in building scalable infrastructure to obtain this information. And then, of course, we would like to exchange the information between all the various places, since they are all linked together.

(Click on image for a larger version)

The vision we have for the Water Resources Observation Network is that we are building a national-scale network which will provide us with information systems which will enable dynamic and timely reporting of Australia's water resources, done through standard interfaces. We are looking to develop new tools to discover resources and also for planning, reporting and forecasting. And with this – this is the interesting bit for an IT group such as the CSIRO ICT Centre – there are many whole new challenges here. There is new equipment that has to be built, along with new information systems that don't currently exist. This has actually created a research direction for even the IT people – it is not as if you can go down to Dick Smith and buy these things off the shelf and then deploy them. It's not like that yet, but we would like it to be like that, where people could easily deploy such systems and be able to data feed into the national system. This is what we are talking about.

And so we are planning to build this system.

(Click on image for a larger version)

Interestingly, we do have a common interface system that exists and everyone has more or less agreed on. That is the internet. Web services are a way of publishing data in a recognised format. CSIRO is a member of the World Wide Web Consortium – the people who created the internet. There are 16 offices around the world and CSIRO hosts one of the offices. CSIRO has an input into setting these standards in an open and collaborative way, and we are looking to making water standards be integrated into the World Wide Web (WWW). This way everyone has a common way of talking about water and the associated data, it allows us to move the data around, or publish the data on a web page or as a data source which can be consumed by other applications.

The key idea is that these systems will be entirely distributed and linked by the internet. You will be able, for instance, in various parts of the country, get access to satellite data or, say, rainfall data for the nation or even just dam levels in a particular area or soil moistures et cetera on a farm. This will facilitate an understanding of how all this information links in and helps to build common information systems.

So this is the grand vision that we are working towards. CSIRO can't do this on its own, and we are looking for partners through our collaboration funds, working with our state governments and the National Water Commission to take this forward as a grand vision. Certainly this is something that we would very much welcome people in the community helping us with. We have the vision, we want to build the systems, and I think this can be done in a cooperative framework. Using the internet as the framework creates a whole bunch of interesting possibilities – we can progressively add and grow infrastructure.

(Click on image for a larger version)

The WRON is not just the internet – it is actually a set of components that are brought together in a loosely coupled system. The internet unites these things. We are talking about making historical databases that have been sampled over more than the last century – being available online. Implementations of climatic models, hydrology models, will be available online. So would real-time in situ sensors – in other words, you might have a water flow sensor in a particular dam or a river – and you could look at the data in real time to see what's happening.

It's quite interesting that, when we have studied some of the effects of real-time data, we have actually seen very different effects that the scientists hadn't predicted, because a lot of the data is temporally sampled. The system will also integrate interesting data sources – things like satellite imagery. The main thing for government policy, users and scientists is to have the real-time reporting tools available as they are needed – and you can compose these services as you need them.

(Click on image for a larger version)

What are some examples of this? One that I have mentioned already is monitoring real time water access and entitlement – in other words, how much water are you entitled to, and how much have you actually used? Other examples are reporting on water usage online, and also looking at the environmental flows that are occurring, for policy makers to decide, 'Well, this is the environmental situation. We need to increase or decrease water flows to maintain or protect our environments.' Another application is planning based on seasonal outlooks, basically taking current water levels plus forecasts and building them forward in time for predictions.

Another exciting application is dynamic water trading. If you know how much water you have got, and with the climate models you can forecast what is going to happen, you could actually create a water futures market. We are looking at these possibilities very closely. Once you have traded the water, you will need to make sure the person who bought the water – gets the water. So there are some quite interesting aspects to this. We believe this can all be done and underpinned by information and communications technologies infrastructure.

Likewise WRON can be applied to environmental flows, quality accounting – understanding what is actually happening with the water – and also looking at trends and forecasts, predicting whether we do need new dams or not in a more structured way, ensuring planning is not done in haphazard ways.

The last of these examples is flexible reporting. We don't always know what sort of application we want to compose, but there are now IT technologies available – given all these data sources, given these particular requirements – to enable the generation of reports about a particular phenomenon.

(Click on image for a larger version)

What will this require? We do need infrastructure. This is not going to happen overnight, and we really do need to upgrade and enhance our monitoring infrastructure. It is pretty poor in our nation, and this is where I think the federal government and state governments can come together – and they are, and we are hoping they can agree to get the necessary infrastructure put in place.

We also need to have interfaces to existing networks and systems, so we can build these bridges between the islands of data; the communication networks for these sensor networks; and data servers and hosting of these types of facilities.

(Click on image for a larger version)

One of the projects that we have been very involved with is building sensor network platforms. Here is an example of one of these nodes which could be, say, sitting in a river or in a dam, or measuring soil moisture.

This board actually contains a computer. Even though, it is small enough to sit in my fingers – it has all the components of a computer on board. It is not a very powerful computer but it doesn't have to be very powerful. All it needs to do is, say, obtain some data, measure something, and then just transmit it. These little on-board radios that we can transmit up to a kilometre. So you don't have to have these computers sitting very closely together; they can be fairly sparsely separated. They form ad hoc networks: meaning that they talk to each other and they transmit data amongst themselves.

If you distribute these component systems with a bit of redundancy, you can make the whole system fairly reliable. So if one link has gone down or is broken – someone trod on it, it got washed away in a flood, something like that – you have got other ways to transmit to get over the problem.

(Click on image for a larger version)

As I have said, these are very small (60 x 60 mm); they use very little power; they have long-range radios; they can self-organise themselves; and they can also talk to a whole bunch of different sensors.

(Click on image for a larger version)

We have also been looking at putting cameras on these devices, to allow images to be captured. You might say that that would be very expensive exercise if you had to transmit all the images. As you have probably seen, the live streaming of data on the internet is very expensive. But, for instance, if we could use a camera to understand flood water levels. Rather than having very complicated instrumentation in the river, we could have a stick there marked up with numbers, and all we would need to do is have a computer vision program – which we have done – enable the camera to read the numbers – allowing you to interpret how deep the water is. The radio would then send off a number – say 4.3 m, which represents water depth. You wouldn't need to do a lot of complicated things with intelligent sensors. (In any case, sensors can get washed away, et cetera.)

We could also estimate things such as water flow by using vision systems.

(Click on image for a larger version)

One of the first systems we have really started to commission is in the Burdekin in Queensland. This is a farming application and we are working with cane farmers. Interestingly, they have been using bore water to irrigate their farms. But they have used the water to the point, when pumping from the aquifers, when there is water now coming in from the sea – which is introducing salinity into the aquifer. Of course, this is not a good thing.

What we are looking at here is having our sensors embedded in the bore, measuring the salinity in real time, and being able to say, 'Well, you can pump water,' or 'You can't.' By controlling how much water is allowed to be used, we can maintain the purity of our water resources.

That data is then shipped off elsewhere and can be accessed – what is the state of the aquifers. This is one of the first commissionings we have done in North Queensland, and the system is being linked in to the whole demonstration system that we are working on.

(Click on image for a larger version)

We have a number of interesting websites. If you have some time and access to a web browser, you can have a look at www.sensornets.csiro.au/qcat, which has interfaces to Google Earth, et cetera. You can have a look, in real time, at what values our sensors are reading and what sort of data is being supplied.

This is soil moisture, water quality, et cetera. Some of these networks have been in operation for nearly two years, continually. We have been working very hard on making these systems easy to deploy and self-maintaining. You don't want someone having to wander around to change batteries, reboot the computers or load new software. The technology must be easily deployed, practically this means being placed and then left alone.

(Click on image for a larger version)

The other interesting website is www.wron.net.au, where you can see a snapshot of the WRON program. It also has some interesting information available online. For instance, it has access to the current levels of all Australian dams. At the moment, there are no real sensors there, but the system is collecting dam levels from other data sources and aggregating them all on one Google-like interface. It is very well done. The work was done by our colleagues in the Division of Land and Water. We are now about to start streaming sensor data onto this website, so you will be able to see what is happening at the Burdekin: what is the salinity in the aquifers? Should the cane farmers be able to pump water?