Millions of Gallons

Water conservation through data analysis

Your water meter already knows where the waste is.

Digital water meters record a reading every 15 minutes. Almost nobody looks at it. We built open-source software that turns that data into a conservation report: leaks, watering in the rain, off-day and midday irrigation, and how much water and money each one is costing.

What four homes' data revealed

Eight months of 15-minute readings from Jericho Water District on Long Island, cross-referenced with rainfall and the district's watering rules.

68 days
of nonstop flow

One home's meter never stopped moving from New Year's Day to March 8. A winter leak, at least 21,000 gallons, found only because the data showed it.

1,500 gal
watered during a tropical storm

On July 10, 2020, Tropical Storm Fay dropped 1.5 inches on Long Island. One sprinkler system ran its full cycle anyway. It did the same on ten other rainy days.

53%
of irrigation between 10 AM and 4 PM

Another home watered through the hottest hours, in sessions of up to 4,400 gallons, when the district prohibits it and evaporation is highest.

0
homes on an odd/even schedule

None of the sprinkler systems followed the district's odd/even calendar. Their controllers run a daily program, not the rules.

The story

In the summer of 2020 two brothers on Long Island watched a neighbor's sprinklers run in the middle of a rainstorm. The water district had rules against exactly that. Nobody was checking, because nobody could: the only people with the meter data were the district and the meter vendor, and neither would share it.

We got a handful of anonymized sample accounts. That was enough to prove the point. This site is the result: the analysis code is open, the method is documented, and the analyzer runs in your browser so you can point it at your own data without handing it to anyone.

Water districts and vendors will not give researchers data. Homeowners can. If enough people run their own analysis and choose to contribute the results, we will finally have a real dataset for studying residential conservation.

What the analyzer checks

  • Leak detection
    Flow that never reaches zero overnight, for days at a time.
  • Watering in the rain
    Irrigation on days with measurable rain, matched to local weather history.
  • Wrong day
    Odd/even, day-of-week, and zone schedules from your district's rules.
  • Wrong hours
    Midday watering that evaporates before it reaches the roots.
  • Overwatering
    Gallons applied versus what the lawn needed, from evapotranspiration.
  • Vacation mode
    Days when nobody was home, and what still ran while you were away.
  • Off-season irrigation
    Sprinklers running when the district says the system should be winterized.
  • Pool fills and anomalies
    One-off spikes that are not the irrigation schedule.

Rules find the events. A decision model gives a second opinion on ambiguous ones, with probabilities. Read the methodology.

Three ways to help

Run your own data

If your utility uses smart meters you can usually export 15-minute readings yourself. Here is how.

Add your district's rules

Every district has its own watering calendar. We are building the first open database of them. Submit yours.

Contribute to research

After an analysis, share the daily totals and findings, anonymously. What we do with it.