Phase 2 audit
The data this model needs
mostly isn’t published
Bengaluru is not short of water data. It is short of water data at the resolution a neighbourhood model runs on. The audit graded every variable the index requires against the resolution it actually needs — and consumption, losses and supply reliability all came back at the bottom of the scale. That finding is the reason this prototype exists, so it is shown first rather than buried in a limitations section.
Grades A–C
4
variables usable at or aggregable to ward level
Grade D
1
categorical evidence only — vulnerability, not measurement
Grade E
3
not available at the resolution the model requires
Availability by variable
Select a row to read what stopped the variable reaching a usable grade.
Indicators feeding the index
- Groundwater depletionCSynthetichigher = more stress
How far the local aquifer has drawn down relative to the rest of the city.
Audit: station observations exist, but reviewed metadata do not identify Bengaluru Urban stations well enough to assign them to wards.
- Supply reliabilityESyntheticreversed
Share of planned piped supply hours a ward actually receives.
Audit: no systematic ward or DMA record of supply hours, interruption frequency or pressure was identified.
- Non-revenue waterESynthetichigher = more stress
Water put into the network that never reaches a billed connection.
Audit: only city-level loss evidence was identified; no ward or DMA measurements are published.
- Demand pressureESynthetichigher = more stress
Demand per connection measured against the ward's available supply.
Audit: no sufficiently usable public ward or DMA consumption series was identified.
- Rainfall deficitChigher = more stress
Shortfall against the 1991–2020 long-period average of 1,077.1 mm.
City-level series is real (IMD, republished by OpenCity). The ward-level split shown here is synthetic: Bengaluru shows low spatial rainfall variability, which limits ward discrimination.
- Tanker dependenceDSynthetichigher = more stress
How far households fall back on tankers and private borewells.
Audit: vulnerability evidence exists as a categorical critical-ward notification, not as a measured quantity.
The one real result
Rainfall is the only variable in the paper with a complete, reproducible public series. Across 33 complete years (1991–2023), annual totals ranged from 587.2 mm in 1994 to 1,957.7 mm in 2022. 18 of those years fell below the 1,077.1 mm long-period average.
It is a single city series. It supports a climate-context indicator; it cannot rank one ward against another.
Generated inputs
The ward lattice and every grade-E series are produced by a seeded generator so any figure on this site can be reproduced exactly and discarded just as easily.
- Seed
- 20300322
- Generator
- gen_wards.py@1.0.0
- Wards
- 36
- Geometry
- Schematic, WGS84
Sources
| Dataset | Publisher | Resolution | Limitation |
|---|---|---|---|
| Bengaluru monthly rainfall1901–2024 | IMD, republished by OpenCity | Single city series | Not ward or DMA resolved; 2024 incomplete. |
| Bengaluru Urban rainfall stations2024 | KSNDMC, republished by OpenCity | Station / taluk | Not joined to verified ward boundaries. |
| Supply-line geometry2025 | Utility source, republished by OpenCity | Line map, diameter classes | No published DMA flow, pressure, demand or reliability series. |
| Groundwater monitoringPeriodic | Central Ground Water Board | Monitoring wells / stations | No consistent ward crosswalk in reviewed metadata. |
| Ward boundary resources2022–2026 | Municipal / metropolitan authority | Ward polygons | Ward schemes change between releases and no matching operational observations exist. |
| Ward operational seriesSynthetic2021–2026 (generated) | JalDrishti synthetic generator | Ward, monthly | Fabricated under a documented seed. Carries no evidential weight and must not be read as a measurement. |
| Representative distribution networkSyntheticStatic topology | JalDrishti synthetic generator | Network zone | A stand-in topology. Not a calibrated Bengaluru DMA and not an EPANET solver run. |