AquaRisk

Inspiration

AquaRisk was inspired by how hard it is for everyday people to understand whether their water may be contaminated by microplastics and chemical waste pollution. There is a huge amount of environmental data available, but most of it is buried in technical datasets and difficult for communities and families to interpret. We wanted to build a tool that makes water contamination risk easier to see, understand, and act on.

What it does

AquaRisk helps people explore how contaminated their water may be by microplastics and chemical pollution. It analyzes industrial and stream linked environmental data to identify facilities and waterways associated with higher contamination risk, then displays that information through an interactive map and facility-level explanations. Users can inspect hotspots and better understand why a certain location may face elevated water-quality risk. Users can also explore predicted estimations of the water quality in the future. The current project environment is confined to the state of Connecticut.

How we built it

We built AquaRisk by processing environmental facility and toxicity datasets, filtering them for Connecticut, and linking facilities to stream segments using geographic identifiers and coordinates. We created a scoring pipeline that aggregates toxicity-weighted concentrations into stream-level risk scores, which serve as a proxy for water contamination risk from pollutants such as heavy metals and microplastic-related industrial sources. On top of that data layer, we built an interactive Streamlit and Folium interface with maps, filters, rankings, and detailed explanations. We also generated fine-tuning data and trained a local model workflow so users can ask natural-language questions about contamination risk and future scenarios. For the predictive side, we gathered a broad set of aggregate historical datasets, identified usable year-to-year patterns, and used those trends to build models that project contamination risk into upcoming years.

Challenges we ran into

One of the biggest challenges was translating messy environmental data into a clear picture of how contaminated someone’s water might be. We had to clean multiple datasets, handle missing geographic information, and connect facilities to segments. Additionally, we had to find a plethora of aggregate datasets that described a clean pattern of the previous years to be able to create a predictive model for the upcoming years. We also had to make the interface intuitive while keeping the analysis grounded in real data.

Accomplishments that we're proud of

We’re proud that AquaRisk makes microplastic and heavy metal pollution risk more understandable for non-experts. We were able to take contamination data trapped in technical files, we turned it into an interactive tool that helps people see local hotspots, explore likely sources, and understand where water contamination risk may be highest. We’re especially proud of the predictive modeling component: we had to find a plethora of aggregate datasets that showed clean historical patterns, use those patterns to understand prior-year trends, and then build a model that could project risk into future years. Bringing together data engineering, visualization, AI-assisted explanation, and forward-looking modeling into one system is something we’re really proud of.

What we learned

We learned a lot about environmental data engineering, geospatial analysis, and how to model water contamination risk from incomplete proxy data. We also learned how difficult and important it is to find historical datasets that are clean enough for predictive modeling, especially when trying to forecast future microplastic and heavy metal pollution risk. On the technical side, we learned how to build a full pipeline from raw data to interactive insights, AI-powered question answering, and future-oriented risk projections.

What's next for AquaRisk

We plan to expand beyond Connecticut, incorporate more direct water-quality and monitoring datasets, and improve the risk model so it captures contamination more precisely. We also want to make the predictive modeling more robust by adding more historical data, refining our forecasting methods, and giving users clearer explanations of how contamination risk may change over time. Ultimately, we want AquaRisk to become a stronger planning and awareness tool for communities.

Built With

• python