University of Wisconsin–Madison

On Lake Forecasts

And Why Limnologists Can’t Be Everywhere at Once – Q & A with CFL Distinguished Professor of Research, Paul Hanson

by Adam Hinterthuer, Limnology News – Number 25, Fall 2016

Here at the CFL, we dream of a future where we’re able to predict harmful algal blooms in the waters right outside our lab’s window and alert the public to unsafe conditions before they occur.  Unfortunately, says CFL’s research professor, Paul Hanson, we’re still a long way off. But, he says, collecting data through a combination of collaborative research, automated buoys and citizen-scientists has the potential to get us closer to that future.

Paul Hanson on the remote sensing buoy on Trout Lake. Photo courtesy: Jeff Miller, UW-Madison Communications

We talked with Paul about lake forecasting, big data, and how scientists are working to get more information more frequently from more places.

So it’s safe to say we’re not going to be predicting algae blooms any time soon?

I think most ecologists are willing to admit that predicting phytoplankton [algae] blooms is extremely difficult, perhaps impossible. Being able to say with precision that “next week we expect something bad to happen” is what ecologists have kind of given up on at this point.

What we feel like we still can do, is to try to determine what’s the probability of something bad happening [to water quality] across different time scales. We might not be able to tell you we know the exact day a bloom will happen during the summer, but we might be able to give you the probability that there will be a bad event.

That sounds sort of like meteorologists forecasting the weather.

It’s a lot like weather forecasting, but even more difficult because of the biological element to it!

What do we need, then, to make better forecasts?

To do a better job of forecasting, we need really good data and we need really good models. So, we have high frequency data coming off the buoy [anchored in the middle of Lake Mendota each season] but that represents just one part of the lake and we’ve done research here at the CFL to show that, at least in the broader timescales like weeks and seasons, data from the middle of the lake gives you a different picture than data from the edges of a lake. (Note: This work was part of CFL alumna, Amanda Stone’s (MS 2012, Carpenter & Hanson), thesis.)

You know, if you go look at our pier out front, you might have a different view than if you were over at one of the beaches. Or if you look at data from beach closures, not all the beaches close at the same time. So there is spatial variability of water quality along the shoreline. So to improve data, we really need more observations in more places.

Obviously limnologists can’t be everywhere at once and buoys are expensive. How do we get that data?

Well, it’s difficult to deploy our really expensive equipment to lots of different places. We just don’t have the money or the resources or people power.  So that’s where some citizen science and water quality apps can come into play.

There are two apps that I’m aware of – one is the NTL Lake Conditions app that was developed here by two students working for Corinna Gries. But that app is about conveying the Lake Mendota data we’ve already collected to the general public.

The one I’m aware of that collects data from citizen scientists is called the Lake Observer App and was developed by Kathie Weathers’ group at the Cary Institute of Ecosystem Studies. That app asks users to look at water quality in the lake or river they’re on and submit observations. There’s also the water quality measurements made around Madison’s lakes by Clean Lakes Alliance volunteers for their lake forecast program.

Citizen engagement and mobile technology is really coming together to provide more data on water quality for lakes. And those data are sort of collecting and queueing up and providing this important repository that we’ll be able to go back to and mine. What is certainly true about ecosystems is that they are highly variable, complex, and difficult to predict. So the data that’s been collected will be put to really great use, because we need tons of it.