AquaSentinel

Turning citizen observations into actionable freshwater intelligence.

Comment

Inspiration

Freshwater ecosystems can change before those changes are formally detected. A citizen may notice unusual water colour, floating debris, changes in wildlife, foam, sheen, or other visible conditions and report them, but a single observation is often difficult to interpret on its own.

We wanted to explore a simple question:

What if citizen observations could become early environmental intelligence rather than simply being stored as isolated reports?

That idea inspired AquaSentinel - an AI-assisted platform that connects citizen observations across time, geography, and evidence to identify potential environmental patterns and help direct human attention toward observations that may deserve further investigation.

Our goal was not to build an AI system that claims to diagnose pollution or replace environmental experts. Instead, we wanted to build a system that can recognize potential patterns, explain why they matter, identify what evidence is still missing, and keep humans involved in the final decision.

What it does

AquaSentinel turns citizen freshwater observations into explainable environmental intelligence.

A citizen can submit:

  • A real geographic location
  • The observation date and time
  • Environmental signals such as unusual colour, strong odour, reduced wildlife, foam or sheen, fish distress, or debris
  • A written description
  • Optional photo evidence

AquaSentinel then evaluates the available evidence and context and produces an assessment:

  • NORMAL
  • WATCH
  • INVESTIGATE

Rather than presenting a single unexplained AI score, AquaSentinel shows the evidence behind the assessment, what information is still missing, and what action may be appropriate next.

For example, during our demonstration:

61% confidence → WATCH

After confirming that the condition had been observed repeatedly:

69% confidence → WATCH

After confirming similar signs at nearby observations:

81% confidence → INVESTIGATE

This demonstrates how additional evidence can change an assessment in an understandable and traceable way.

AquaSentinel also connects observations across geographic and temporal context, provides a historical timeline, identifies correlated observations, and visualizes potential patterns through a resilience map.

The platform keeps a human-in-the-loop throughout the process. An assessment is a decision-support signal, not a scientific finding.

“AquaSentinel explicitly identifies missing evidence such as field measurements or laboratory results and supports human verification before environmental action.”

How we built it

We built AquaSentinel as a responsive web application with separate components for citizen observation, evidence and context analysis, assessment, persistence, and visualization.

The core architecture is:

Citizen
   ↓
Observation Capture
(Location • Time • Signals • Photo)
   ↓
Evidence & Context
(Current + Historical)
(Geographic + Temporal)
   ↓
AI Investigation
(Follow-up Questions)
   ↓
Assessment
(Normal / Watch / Investigate)
   ↓
Human Verification
   ↓
Resilience Intelligence

The application supports real location selection and real observation timestamps instead of relying only on predefined demonstration locations.

To demonstrate historical correlation without presenting fabricated environmental measurements as real data, we included clearly labelled synthetic demonstration observations. These records allow the prototype to demonstrate how patterns can develop over time while making it clear that the historical records are not real environmental measurements.

The assessment system was designed to be explainable for the prototype. Follow-up evidence changes the assessment, allowing users to see how additional information affects the result rather than receiving an unexplained prediction.

We also designed the interface to work across desktop and mobile-sized screens, making citizen reporting accessible from different devices.

Challenges we ran into

One of our biggest challenges was balancing AI assistance with scientific responsibility.

It would be easy to build a system that sees unusual water colour in an image and immediately declares "pollution." However, a visual observation alone cannot establish a cause.

We therefore designed AquaSentinel around uncertainty.

The system distinguishes between:

  • What was actually observed
  • What evidence supports a potential pattern
  • What evidence is still missing
  • What the system can reasonably infer
  • What requires human verification

Another challenge was demonstrating geographic and historical correlation without misleading users. We solved this by clearly identifying historical records as synthetic demonstration data, while allowing new observations to use actual selected locations and timestamps.

We also wanted the assessment to be understandable to people without specialized environmental expertise. Instead of hiding the reasoning behind a single score, AquaSentinel exposes supporting evidence and missing evidence and shows how additional information changes the assessment.

Finally, we had to prioritize the most meaningful capabilities within the hackathon timeframe. Rather than building many disconnected features, we focused on creating a complete working prototype that demonstrates the central idea from citizen observation to environmental intelligence.

Accomplishments that we're proud of

We are most proud of turning a simple citizen reporting concept into a system that can connect observations, evaluate evidence, explain uncertainty, and identify potential environmental patterns. The assessment progression is one of the features we are particularly proud of:

61% WATCH → 69% WATCH → 81% INVESTIGATE

The system does not simply increase a number. Additional evidence contributes to a change in the assessment, and the user can see why.

We are also proud of the human-in-the-loop design. AquaSentinel deliberately avoids presenting prototype assessments as confirmed pollution events, diagnoses, or proof of environmental causality.

The project naturally connects several OneAquaHealth themes:

  • AI-Supported Assessment
  • Data-to-Insight
  • Resilience Informatics

We are also proud that the prototype combines citizen observations with geographic and temporal context, historical patterns, explainable assessment, and resilience visualization in one accessible interface.

What we learned

One of the biggest lessons we learned is that responsible environmental AI is not only about making predictions more accurate.

The evidence behind a prediction and the transparency of the decision are equally important. We learned the value of asking targeted follow-up questions. A single citizen report may contain limited information, but carefully selected questions can provide additional context that changes how an observation should be assessed.

We also learned that uncertainty should be visible rather than hidden. Instead of saying:

"This is pollution."

A responsible system should be able to say: "This is a potential anomaly. Here is the evidence supporting that assessment, here is what is missing, and here is why human verification may be needed."

That principle became central to AquaSentinel.

What's next for AquaSentinel

The current prototype demonstrates how citizen observations can be transformed into explainable environmental intelligence. The next step would be connecting AquaSentinel to richer real-world environmental data.

Potential integrations include:

  • Citizen-science platforms
  • Environmental sensors
  • Weather and rainfall data
  • Hydrological data
  • Biodiversity observations
  • Verified laboratory measurements
  • Standardized environmental and health datasets
  • Municipal and environmental response systems

With these additional data sources, AquaSentinel could become a broader early-warning and environmental decision-support platform for freshwater ecosystems.

The long-term vision is not to replace scientists, environmental agencies, or citizen-science platforms.

It is to help connect their data and observations so that potential environmental changes can be identified earlier, understood more clearly, and reviewed by the right people.

AquaSentinel turns isolated citizen observations into explainable environmental intelligence — helping transform observations into actionable insights while keeping humans at the center of environmental decision-making.

Built With

Share this project:

Updates

Submission history