Backcountry Water Oracle
33.96905, -111.50807

Upper McFarland @ #88/95

33.96905, -111.50807

Long unreportedlast reported 9 years ago

How this site decides what it can and cannot say — the method.

Insufficient record

No read issued

2 reports — a correlation needs at least 10 before it means anything, so no read is issued here.

What people saw

2 observations between 2017-06-11 and 2017-11-18, every one pools or dripping. The most recent was 9 years ago.

These are the observations themselves, not a forecast. Too few to correlate against rainfall, but they are what people actually found.

  • 2017-11-18Pools or dripping0.2Two large pools with clean, clear water.
  • 2017-06-11Pools or dripping0.2Pools below junction.

Why

The engine will still produce a number from 2 reports, and that number is noise. It will happily report a perfect correlation from two observations. On a screen where someone is deciding how much water to carry, that is the most dangerous thing this site could render, so it is not rendered.

8 more reports would reach the 10 needed for a read.

This is the fix

Every dated observation moves this source closer to a read, and sharpens the ones around it too.

Add a report

What was the water doing when you were there? Every observation sharpens this source and, because nearby sources lend each other statistical strength, the ones around it.

The day you were standing there, not today. The forecast correlates rainfall up to this date, so a wrong date is worse than no report.

How much water?

Score the water you could actually use. If the seep itself has stopped but the rock tanks below it are holding, that is water — score the tanks.

In your own words. Kept verbatim — the number is what the model reads, but this is what a human checks when a forecast looks surprising.