The digital transformation of water and wastewater utilities may be early-stage, but it is making tremendous impact where applied, and most organizations already have the data at their disposal to improve their local water conditions.
This has been realized at DC Water, where artificial intelligence (AI) and machine learning (ML) are being used to save the Anacostia River, and also in Sierra Leone, Africa, where it is being used to save lives. In both cases, but in different ways, predictive modeling facilitated by DataRobot’s AI platform was used to inform decision-making, optimize operations, and preserve water.
For this Q&A, I spoke with DataRobot’s Paul Fornia about how AI is working for the above-mentioned projects – and how it can be applied to all sorts of water scenarios – powered by technology and information that is often available but largely underutilized.
What are the conditions of the Anacostia River in terms of current water quality, how it is trending, and the factors affecting it?
Historically, the Anacostia River has seen high levels of pollution. While there has been improvement over the last few decades as more money and effort has gone into restoration efforts, there was still a pollution concern for both the river as well as the upstream on larger bodies of water, such as Chesapeake Bay and the Atlantic Ocean.
Similar to many water sources, the Anacostia River saw bacteria levels fluctuating every day, which were contributed by various factors such as rain, turbidity, water levels, the tide, pH of the water, and sunshine.
What steps have been taken to improve water quality?
Previously, there have been several initiatives to improve water quality, such as the Clean Rivers Project and green infrastructure projects around DC that prevent runoff. Today, there is a dedicated team of volunteer citizen scientists that monitor several sites in the area with manual water samples, about once per week. The volunteer water quality monitoring, named DC Volunteer Water Quality Monitoring, is a program funded by the District Department of Energy and Environment to monitor water quality changes. The results of these tests are posted to a public-facing website and provide a valuable window into water quality for the region. Unfortunately, because of the time-intensive testing process, these measurements are only provided on a weekly basis, and even the latest measurements will be about 24 hours out of date. This is especially problematic if bacteria levels are quickly changing. For example, if there was a big rainstorm that possibly caused sewage to overflow into the river, the team wouldn’t have the results quickly enough to share with the public before the water quality was altered.
How did DataRobot get involved in the effort?
DataRobot’s work with the Anacostia Riverkeeper is the result of our AI for Good: Powered by DataRobot program, in which we donate our enterprise AI platform to social good organizations to solve critical, global challenges. We also allocate data scientists and dedicated success resources to ensure the recipients are able to see the long-term benefits of our AI technology. Anacostia Riverkeeper applied to our program in order to augment their water quality monitoring processes in hopes to better predict bacteria levels based on known current conditions and past information.
What is the technology deployed, and how does it work?
The Anacostia Riverkeeper uses DataRobot’s enterprise AI platform, which automates the entire end-to-end AI lifecycle allowing them to build, deploy, and manage machine-learning models. In the Anacostia Riverkeeper’s use case, their "citizen scientists" collect the data from their monitoring processes and input that data into DataRobot’s AI platform where it can then predict the current state of water faster and with more accuracy. The platform will also use and learn from supplemental public data sources like USGS and NOAA.
How do forecasts help improve water quality?
While this project is still in its early stages, we’re expecting our platform to supplement the normal process of water-quality testing. By speeding up these processes, the river can be maintained more efficiently and more frequently, giving policymakers and the public even more transparency into the water quality of the river.
How may this project influence future projects elsewhere?
One of the additional goals we have set for this project is to expand to other sites such as the Potomac River and Rock Creek. One of the key features our platform is its ability to scale, so we’re hoping to help other urban riverkeepers in other regions to accurately test and improve their water quality given it’s such an issue globally.
DataRobot also has a hand in other water-related projects, such as our work with the Global Water Challenge (GWC). Through our partnership with GWC, we’ve analyzed more than 500,000 data points to predict waterpoint breaks in African countries in order to ensure all people have regular access to water. Working with GWC, local governments in Africa have been able to proactively identify at-risk waterpoints and better manage water-related budgets and programs for repairs and new construction. The program is up and running in both Sierra Leone and Liberia — both of which have been so successful that leveraging DataRobot for improved water access is now being mandated across Africa. Not only has this project improved population health, but it has also significantly cut down individuals’ time spent searching for water, and has allowed people to resume daily tasks, like attending school and their jobs.
What are the future plans for DataRobot?
Through our AI for Good program, we plan to continue using our AI platform to solve some of the greatest challenges facing society today. We firmly believe that AI has the power to make a meaningful difference, and we will continue to share our platform with organizations to enable sustainable, AI-driven solutions that drive that change and better the world. To expand this program even further, we accept applications on a rolling basis and are always looking for new companies to make a profound impact with.
Paul Fornia is a data scientist at DataRobot, where he helps enable organizations from a variety of industries (with a focus on non-profit partners) to build AI systems. He also has several years of experience working and consulting in the federal sector. Paul studied math and economics at the University of Colorado and will be completing his master’s degree in Computer Science from Johns Hopkins this summer.