Posting in Education
With a project that required measuring water levels -- and without the thousands of dollars needed to set up a sophisticated system -- a researcher turned to crowdsourcing.
With a project that required measuring water levels -- and without the thousands of dollars needed to set up a sophisticated system -- a researcher turned to crowdsourcing. The result is CrowdHydrology, a project that uses text messages from citizen scientists to collect water level data.
What is CrowdHydrology?
It is a mini-research project looking at how water levels in a watershed change through time and the seasons. One of my graduate students is doing studies of ground water-surface water interaction in a region of western New York that has no monitoring stations for the level of stream. We needed to find a way to figure out how water levels are changing in the stream. [That way] we can have an idea when we're looking at ground water-surface water interaction if this pulse is [due to] storm runoff or something else. We wanted to identify where these pulses were coming from in the watershed and we needed a cheap way of doing that.
The CrowdHydrology idea is to put these simple gauging staffs in the stream with a sign [asking passers-by to send] a text message telling us the water level. This allows us to then get some idea -- little snapshots -- of what the water levels are doing within a larger watershed.
What inspired you to use crowdsourcing to monitor water levels?
I read an article, I think in the Wall Street Journal, about using smart phones for science. There's a researcher at UC Davis that uses smart phones to identify road kill. He's interested in wildlife corridors and identifying where animals have been hit by vehicles. When I read this article, I said: If someone is willing to send a picture of road kill, there's no way I can't get someone to send me a text message of the stream stage.
In terms of an economic side, the [U.S. Geological Survey] has gauging stations that are unbelievable. They measure water levels every hour with a pressure transducer. You can go online and see the stream stage real-time. But those cost about $15,000. I need water levels over a larger area and I don't have $15,000 for every one of these sites. Using crowdsourcing, I can build one of these stations for about $60.
Why is it important to monitor water levels?
We're interested in this mixing zone between ground water and surface water in streams. Some of the constituents that move through these zones are things like nitrates. You can have these pulses of nitrogen coming off farm fields going into streams where you have large rain events with a lot of runoff. We'd like to know: Are the levels we're measuring a result of this large pulse of overland flow flushing this stuff to stream? Or is something happening on a larger flow path as a result of ground water transporting this over the course of years? These gauges tell us when this stream pulse may be coming through the watershed. In making these measurements, we can see when precipitation occurred, when we see this pulse move through the watershed and what we're monitoring in terms of groundwater.
Talk about your pilot site. Describe the system there and talk about how much participation you've had so far.
The first site we set up in the middle of May 2011. It's at the Buffalo Audubon Society's site called Beaver Meadow. It's a wetland environment where there's an open water pond. At the outlet to that pond, I've installed a vertical piece of wood with a large ruler attached to it. At the top of the ruler is a sign that says, "Text water levels to..." There's a phone number and a station number. That's a meter and a half into the water, right off a hiking path. On the hiking path, I have an informational sign that explains to people what we'd like them to do and why we'd like them to measure the water levels in the pond.
In terms of interest at the site, we may be up to 55 measurements. Those are 55 text messages of what the water level is in the pond. The text messages were coming in about every day, but there are occasionally lulls.
What are the limitations of this technology?
You only get a text message if someone is willing to share that information with you. It's a very small portion of the community that's actually sending me these text messages at this point.
You're asking a citizen scientist to send you this message. There is some quality control that comes in. There have been a few text messages that made me say, "That's not the right water level."
What's the next step?
I just installed eight new sites. I have a total of nine sites in western New York. Ideally, my vision is that we can run the web site of this at the University at Buffalo. Then anyone in the country who wants to set up one of these sites can set it up. We can send them a unique station number. People can text the data and we can keep it on our website for them. We can be the stream stage hosting institution. If you're in Kansas and you're a middle school teacher that wants to have your class do this, it costs $60 to set this up. It's a great tool for people interested in K-12 education and also for researchers who have a site they can't afford spending $15,000 on.
Contact Lowry at CrowdHydrology.org
Photo, top: The CrowdHydrology gaging staff in action
Aug 15, 2011
Quality control of the data is a major problem with a project like this. Ensuring gathering methods are consistent becomes a problem with the more people you have involved in the data gathering. Minor inconsistencies in collection methods are magnified when you have several people or several dozen people all adding in their own minor errors to the collected data. Maintaining data integrity is tough enough when dealing with a coordinated team of trained people. To add in the variable of random self proclaimed citizen scientists helping is asking for trouble when doing serious science. Cost effective, maybe, but while it might be an acceptable data collection method for a high school science project, I would not trust the data for major decisions.
Unless there is some built in systemic error the accuracy of combined data tends to improve as the number of measurements goes up and random errors tend to cancel each other out. Regarding the specific research mentioned in the article it doesn't sound as though the exact water level is particularly important. They just need an indication of when the water level changes significantly so they can compare it with changes in the chemical makeup of the water and weather events.