Field notes: the district where the register listed boreholes that weren't there

There's a particular kind of silence that happens when a field team pulls up to a GPS point marked "borehole, functional" on a district WASH register and finds a dry patch of ground. No headworks, no apron, no rusted handpump stub. Just grass, and the faint depression where a headworks used to sit, or never did at all.

I've heard some version of this story from enough programme officers now that it stops sounding like bad luck and starts sounding like a pattern. A register gets built up over a decade, through two or three donor cycles, each one adding points and rarely subtracting them. Contractors self-report completion. A supervisor signs off from a desk two hundred kilometres away. Somewhere in year four, a handpump gets cannibalised for parts, or a community relocates the water point twenty metres over for drainage reasons nobody wrote down, or the whole thing was logged from a work order that got amended after the fact and the register never caught up.

None of this is fraud, usually. It's just what happens to a spreadsheet that outlives the people who built it.

The district I'm thinking of had 340-odd points on the books. Coverage maps built from that register looked reasonable, even good, in the sense that the dots were spread evenly enough to pass a glance. Planning a WASH intervention off that register meant you'd send a drilling team to top up supply in a ward that, on paper, already had six working boreholes per the register. Three of those six turned out to be phantom points. One was a borehole, but for goats, not people, sunk by a private farm a kilometre outside the settlement boundary the register had assigned it to.

What struck me wasn't the error rate. Registers drift, everyone in this line of work knows that. What struck me was how long it took to find out, and how much of a field team's week got burned confirming it. Ground-truthing 340 points by vehicle and on foot, cross-referencing against GPS that was itself sometimes off by a ward, is not a two-day job. It's the kind of task that eats the first month of a project before anyone's touched a borehole.

Looking at the ground from above before sending a vehicle out answers the easy half of that question fast. Ground-truthing still has to happen, since a handpump handle needs turning and a yield needs checking, and no pixel can do either. Running an overhead pass first narrows that list down to a fraction of itself before a single vehicle leaves the yard. A single clear image of the district shows within a day which of those 340 points actually have a structure sitting on them and which are coordinates attached to nothing. That's the difference between a field team doing confirmation visits and a field team doing discovery visits, and it's the difference between a two-week scoping trip and a two-month one.

This is more or less the gap Sanitation Mapping is built to sit in: an annual imagery pass over a district that turns visible boreholes, tank stands, and latrine blocks into a dated, located point file you can set next to the existing register and see where the two disagree. It won't tell you which borehole still yields water in the dry season. It will tell you, fast, which of the register's phantom points are worth a drive and which ones the field team can cross off before they've even loaded the truck.

The district eventually got its register cleaned up, point by point, the slow way. Most of them do, eventually. The question worth asking before the next programme cycle starts is whether that cleanup happens in month one of fieldwork or gets found out mid-drilling season, when it costs a lot more than a day of someone's time.

If your own register has a few of these phantom points hiding in it, an annual imagery pass is a reasonable place to start sorting out which ones.