Anyone can check the health of their local stream.
StreamSense guides volunteers through a 10 minute stream survey. A photo assistant suggests answers and explains why. You make the final call.
Field record
Example
- Site
- Millbrook, by the footbridge
- Weather
- Rained in the last day
How clear is the water?
Murky, hard to see into
AI: Murky. Brown plume below the pipe.
High confidence. Volunteer agreed.
Can you see any pipes flowing into the stream?
Pipe pouring out smelly water
AI: Pipe pouring out clear water.
Medium confidence. Volunteer chose differently, and their answer counts.
What does it smell like?
Sewage or toilets
Volunteer only. Photos can't show smell.
How it works
You look
Answer 10 simple questions at the stream. Each one explains the science word behind it and why it matters.
The AI suggests
A photo assistant suggests answers and shows how sure it is. Nothing is filled in until you agree.
People check
Fixed rules work out the score and raise health alerts. Unusual reports go to a reviewer before they are published.
Using AI responsibly
- The AI suggests, you decide
- Suggestions are never filled in for you. Your answer always wins, and every choice is saved for reviewers.
- Everything is explained
- Each suggestion shows how confident the AI is and what it saw. Every score shows what lowered it.
- The AI never scores
- The health score and alerts come from fixed, published rules the AI can't change.
- People have the last word
- When the volunteer and the AI disagree, an expert reviewer takes a look.
One stream, three kinds of health
People
Signs of sewage or toxic algae near places where children play or people paddle are flagged as urgent.
Animals
Dead fish, oily films and algae that can poison dogs raise animal health alerts.
Nature
Bank plants, erosion and flow show how well the stream can clean water and support wildlife.