Purpose
The model was built to support one explanatory question: what happens to a litre between rainfall and its return to an estuary? It favours readable system balances over operational detail. Figures have realistic orders of magnitude, but realism is not validation. A real water-resources plan would require observed weather, licensed abstraction limits, asset records, demand forecasts, water-quality sampling and uncertainty analysis.
Reproducible generation
The monthly series is generated by data/generate.mjs with the fixed integer seed 20250814. A linear congruential generator supplies a small deterministic variation around a seasonal rainfall curve. May to August in 2018 and 2022 receive a dry-weather multiplier. Reservoir storage begins at 82 per cent and changes each month by a rainfall contribution minus a seasonal withdrawal. It is bounded between 31 and 96 per cent.
Running the generator again produces the same JSON and CSV byte for byte. The JSON is the source used by the charts. The CSV is a normalized export: each row identifies a series, a period or category, a value and a unit.
Download the full dataset as CSVHow each figure is made
Rainfall and reservoir level
Monthly rainfall is a seasonal curve with seeded variation. The two designated dry summers are multipliers within the generator, not edits made in the chart. Storage is a simplified balance, not a hydrological routing model. It does not represent groundwater, multiple reservoirs, daily storm sequences, changing catchment land cover or forecast-based operation.
The one-litre balance
The balance allocates 3 per cent of raw input to treatment processes, 21 per cent to distribution leakage and 76 per cent to treated water delivered to customers. It is an accounting device. In a real utility, treatment use is removed before water enters distribution, so leakage would ordinarily be reported as a share of distribution input. Here all three values share the same raw-water denominator so the parts visibly total one litre.
District leakage
District rates are invented estimates based on a fictional night-flow analysis and then selected to reconcile approximately with the city balance after allowing for different district volumes. The simple table does not publish those volumes, confidence intervals or the split between mains and customer service-pipe leakage. It should not be used to compare real pipe materials or repair policies.
Household use
Annual litres per person per day follow a constructed declining path, with a 2020 home-working adjustment. The metering marker in 2019 indicates timing, not proof of causation. Appliance standards, price, weather, occupancy and housing mix can all change demand at the same time.
The household calculator
The calculator begins with the 2025 baseline of 136 litres per person per day. It applies a factor of 0.91 for a metered home and 1.04 for an unmetered home, multiplies by household size and 365 days, then rounds to the nearest thousand litres. The comparison median is fixed at 108,000 litres a year.
That deliberately small model ignores dwelling type, income, appliance efficiency, gardens, time away, non-resident visitors and existing plumbing leaks. Meter status is associated with household behaviour but does not determine it. The result is a prompt for scale, not a diagnosis or tariff estimate.
Other system figures
The population is fixed at 412,000. Treatment capacity is 94 megalitres per day; operating energy is 438 kilowatt-hours per megalitre. The return model gives 92 per cent sewer capture, an average treated effluent flow of 48 megalitres a day and 81 per cent nitrogen removal. These values sit in the downloadable data because they appear in the writing, even where they do not have a chart.
What is not known
No probabilistic uncertainty is attached to any number. The monthly storage calculation does not reproduce daily peaks, and the annual wastewater balance cannot describe the ecological effect of a short overflow during a particular tide. The model does not close every household subcategory against billed consumption, trace energy through wastewater treatment, or estimate greenhouse-gas emissions.
Those omissions matter. A system diagram can imply that each boundary is measured cleanly when real boundaries are inferred from meters with error, changing populations and water that takes paths nobody observes. The appropriate reading is structural: rainfall timing affects storage; treatment has material and energy costs; leakage wastes work already done; household demand alters both supply and sewer flows; treated discharge remains part of the estuary.