Skip to content
No. 017On holdWater and sanitation in poor regionsUpdated Nov 15, 2026

Estimating where human waste ends up in cities that have never been surveyed

Good it would do
3 / 5
Can AI help
2 / 5
How neglected
3 / 5
Can we check it
3 / 5

Why it's on hold

Parked: nobody predicts shit-flow-diagram steps for unsurveyed cities and the 405 SFD records are downloadable, but most labels are desk-based defaults (only 55 Comprehensive, 4 in Africa) and the steps that decide the unsafe share (where emptiers dump, whether plants work) are invisible in surveys and satellites, so Solvable is 2. The outside skeptic warns even a narrowed test may only show SFD reproduction.

Why it's stuck

A 2020 analysis of 39 cities with 72 million people found 58 percent of human waste is not safely managed, but each city's estimate (a "shit flow diagram") takes months of local fieldwork and only around 90 such reports existed. The authors found data on emptying, treatment and containment were often missing, so key figures rested on assumptions such as 50 or 100 percent treatment efficiency.

What we'd try

Train models that predict each step of a city's waste flow (contained, emptied, delivered to treatment, treated) from public data that covers every city: household survey questions on toilet type and emptying (DHS and MICS), satellite building density, sewer and treatment plant locations, and groundwater depth. Then produce estimates with uncertainty for every city in low-income countries.

We'd look again if

Both: (1) labels with real provenance: an SFD or JMP expert confirms a usable set of measured (not defaulted) SFD fractions for the narrowed target (contained and emptied shares), including outside South Asia, or DHS-8 emptying modules or JMP SMOSS service-provider data on delivery and treatment become public; (2) a user (JMP, SFD Promotion Initiative, a funder or ministry) names the decision a modelled city estimate would change and its current information source. The first step then is a provenance audit of the 405 SFD records.

Source: https://frontiersin.org/journals/environmental-science/articles/10.3389/fenvs.2020.00001/full

Think we got it wrong?

Ruled-out problems can come back. Tell us what we missed.

Members can read the full record for this problem, check its sources and comment. Open the member view →