How IBM is using smart sensors and computer modeling at Lake George, NY

IoT-based sensors are being used at Lake George, in New York, in order to collect data that can be used to improve the freshwater ecosystem.

How the IBM Multicloud Manager automates how companies manage, move, and integrate apps across clouds

IBM Research is working with partners from the Rensselaer Polytechnic Institute and The Fund for Lake George to place IoT-based sensors around Lake George, in New York, in order to monitor threats such as pollution from road salt, invasive species and excess nutrients.

The Jefferson Project began in 2013, and it includes a smart sensor network that gathers more than 9 terabytes of data annually, and it also uses computer modeling to better understand freshwater ecosystems, and provide science-based insights.

"The motivation to start The Jefferson Project really was human impacts on fresh water, starting in Lake George, and always with the idea that we would expand that as the preservation journeys to other lakes around the region and other lakes around the world, and even other environments, such as marine, streams, rivers, etc.," said Rick Relyea, director of the Jefferson Project and a professor of biological sciences at Rensselaer Polytechnic Institute.

Technology used at Lake George

The lake provided a new way to use technology such as IoT in order to measure data and figure out how humans are negatively impacting the lake.

Some of the technologies, such as hardware and software, were created in-house, between IBM and Rensselaer, and that gave them the opportunity to do something no one else had done before, he explained.

There are 52 multi-sensory platforms at the lake. They're networked mostly through conventional 3G and 4G networking coverage at the lake. The data is brought back into an aggregator and a back end system that handles display technologies and analytics, explained Harry Kolar, associate director of the project, IBM Research.

"We're pushing more and more analytics down to the actual sensor nodes. And the sensors are collecting physical, chemical and biological parameters," Kolar said.

The sensor platforms


IBM and RPI researchers on Lake George as the vertical profiler is being towed to its deployment location.

Image: IBM Research

The sensor platforms consist of five floating, vertical profilers that monitor conditions from the surface of the lake down to the bottom of the five deepest basins.

The IoT nodes are set up to be very intelligent to truly push water quality measurements to create high data quality from the platform. "the platform also is able to do things like adaptive sensing. So if there are changes in the environment, the sensor platform is aware of it from not only its own measurements, but also measurements from other sensor platforms anywhere around the lake," Kolar said.

"It can change how it actually measures. So it may change sampling frequency, if there's a big storm coming, or if there's a particular event that we're expecting that we want to make certain measurements on, the sampling profile would change automatically. All of this is then connected. We actually modeled the entire ecosystem of the lake and the entire watershed," Kolar said.

"We are running computer models of high resolution weather on a daily basis, which then feeds a hydrological model, which is the runoff model, to say where any precipitation ends up, and how it flows over the land surfaces and into the lake itself. And then also those all then feed another computer model. These are all physics-based models, high performance computer models, that then basically model how the water is moving within the lake," Kolar said. "All that's tied together and produces a 36-hour forecast every single day. And the resolution of the weather is at 333 meters. So we have a weather forecast every 333 meters in the entire watershed, every 10 minutes for 36 hours."

This forecast is extremely accurate and it ties to the other models to give a big picture idea of what's going on at the lake compared to what is expected based on the scientific reasons and that is used to further improve the models.

Impact of road salt

Road salt is one of the big issues affecting Lake George. "We say we have the most aggressive comprehensive road salt reduction program, if not in the world, certainly in the country. For an entire watershed, and it involved public and private players in pursuit of reducing the use. And we're working to get the levels of road salt down to 50% of what they currently are by as early as 2020," said Eric Siy, executive director of The FUND for Lake George.

"The approach that we're taking and the linkage to the Jefferson Project technologies and research are really helping to guide where and how we achieve these reductions. And I think most exciting is as we use road salt as an example, we've reduced the use, and we are starting to measure a significant reduction through the smart sensor network. We will actually begin seeing those reductions in real time in the waters, in the streams, in the lake itself. And what does that do? That reinforces our commitment to continuing on this path and showing others the way, leading by example," Siy said.

There's even a mobile weather app that's still in beta testing that can provide hyper-local forecasts for every part of the lake. IBM Research is working on the app so that it will utilize data from the smart sensor network.

What the future holds

The hope is that this smart sensor network, combined with the ability to create models to forecast scenarios, will protect both Lake George and other lakes and bodies of water in the future.

"These are very complex threats facing Lake George and any water body. It takes this level of complexity in addressing the problems to insure we have the best possible chance of protecting this," Siy said.

Also see


A RPI researcher doing some sampling on Lake George.

Image: IBM Research