• Rural water
  • via Proxify

Magenta Insight / Uptime

Reading hand-pump use from the temperature sensor on a well.

The challenge

When people pump water, it flows over a temperature sensor on the well. If ambient and groundwater temperatures differ, this causes a sudden temperature change.

The sensors are read out every few months, so the analysis is offline.

What I did

I built two detection methods. The first extracts local extrema with the slope profile. The second decomposes the signal into trend, seasonality and residual; the residual identifies pump use.

I also built a Dash dashboard for quick visualisation and parameter tuning, and a batch engine that processes many sensor files at once.

Results

A report per pump with the number of pump uses, the longest and shortest use, and the start and stop times of each use.

Both methods agree well with ground truth where ambient and water temperatures differ. When they become similar, detection from temperature alone is not possible, and the dashboard shows this.

In the client's words

“Roel was great. He seamlessly fit into our team and exceeded expectations almost immediately. He's built exactly what we need quickly and effectively.”

Duncan McNicholl, Co-founder & CEO, Uptime Global

Contact

Bring me your analytics question


Send a short email. I reply within two working days with whether I can help, a rough plan and a fixed quote.