Why the number is bigger than people expect
One inch of rain on a 1,000 square foot roof is 623 gallons. Most people guess a tenth of that, because a rain barrel holds 50 gallons and an inch of rain does not look like much. It is worth pausing on where the figure comes from, because it is pure geometry: a square foot under an inch of rain holds one twelfth of a cubic foot, a cubic foot is 7.48 US gallons, and 7.48 divided by 12 is 0.623.
That means a 2,000 square foot roof in a 38 inch climate intercepts about 47,400 gallons a year before any losses — enough to flush a household's toilets several times over, or to irrigate a substantial garden. The gap between that theoretical figure and what a system actually delivers is what this calculator quantifies, and the gap comes from four places.
Runoff coefficient. Not all rain that lands on a roof reaches the gutter. Some wets the surface and evaporates, some blows off, some soaks into a porous material. Smooth impervious surfaces lose least: metal and glazed tile run around 0.90 to 0.95, asphalt shingle around 0.80 to 0.85, and rough or absorptive surfaces considerably less.
Conveyance and filter efficiency. Gutters overshoot in intense rain, leaf screens shed water, vortex filters divert a slice by design. A well-built system holds around 90 percent; an undersized gutter on a steep roof in a thunderstorm can lose far more.
First flush. The first fraction of an inch off a roof carries the accumulated dust, pollen, bird droppings and airborne deposition of the dry period before. Diverting it improves water quality markedly and costs you a fixed depth on every storm — which is why frequent small showers yield disproportionately little.
Tank overflow. This one the calculator does not model, and it is usually the biggest loss of all. Water that arrives when the cistern is already full is gone. Yield is capped by storage and by demand timing, not just by rainfall.
The formula, and the metric version that needs no constant
The US form is V = A × d × 0.6233 × Cr × η. Multiply plan area in square feet by rainfall depth in inches, convert to gallons with the 0.6233 constant, then knock off the two efficiency factors.
The metric version is one of the tidiest identities in applied hydrology: one millimetre of rain on one square metre is exactly one litre. A cubic metre is 1,000 litres, a millimetre is a thousandth of a metre, and the two cancel. So V (litres) = A (m²) × d (mm) × Cr × η, with no conversion constant at all. If you are working in metric, do not introduce a factor — you will only be introducing an error.
Measure the roof in plan, not along the slope. This trips up nearly everyone. Rain falls vertically, so what determines interception is the horizontal footprint the roof covers. A 12:12 pitched roof has 1.41 times the surface area of its footprint and collects exactly the same water as a flat roof of the same footprint. Use the building's outline plus the overhang.
Subtract first flush per event, not per year. The diverter refills and dumps on every storm, so the annual loss is the diversion depth times the number of storms. Fifty events at 0.02 inches each is a full inch of rainfall gone — meaningful in a dry climate, trivial in a wet one. Typical practice sizes the diverter at roughly one to two gallons per 100 square feet of roof, which is 0.016 to 0.032 inches of depth.
Worked example: a 1,500 ft² roof in a 40 inch climate
An asphalt-shingle roof with a 1,500 ft² footprint, a runoff coefficient of 0.85, a well-built conveyance system at 90 percent efficiency, a 0.02 inch first-flush diverter, 40 inches of annual rainfall spread over 50 storms, water and sewer at $6.00 per 1,000 gallons, and 100 gallons a day of garden demand.
- Yield per inch. 1,500 × 0.6233 = 934.95 gallons of raw interception. Apply the runoff coefficient: 934.95 × 0.85 = 794.71. Apply filter efficiency: 794.71 × 0.90 = 715.24 gallons per inch of rain.
- A one-inch storm. Effective depth is 1 − 0.02 = 0.98 in, so 715.24 × 0.98 = 700.9 gallons from a single storm. That is fourteen rain barrels, which is why serious systems use cisterns rather than barrels.
- Annual effective depth. 40 − (0.02 × 50) = 39 inches.
- Annual harvest. 715.24 × 39 = 27,894 gallons, or 105,588 litres.
- Water bill saved. 27,894 ÷ 1,000 × $6.00 = $167.37 a year.
- Days of demand. 27,894 ÷ 100 = 279 days of the garden's requirement — for a year that is 365 days long, so this roof cannot supply that demand year-round even before considering when the rain falls.
That last line is the useful one. The harvest is large, and it is still not enough for the demand as stated, and the shortfall is worse than 86 days because rain and irrigation demand are anti-correlated: you need the most water in the driest month.
Yield per inch of rain by catchment area and surface
| Plan area | Raw (Cr = 1.00) | Metal roof (0.95) | Asphalt shingle (0.85) | Rough or porous (0.70) |
|---|---|---|---|---|
| 500 ft² | 311.7 | 266.5 | 238.4 | 196.3 |
| 1,000 ft² | 623.3 | 533.0 | 476.8 | 392.7 |
| 1,500 ft² | 935.0 | 799.5 | 715.2 | 589.0 |
| 2,000 ft² | 1,246.6 | 1,065.9 | 953.6 | 785.4 |
| 3,000 ft² | 1,869.9 | 1,598.9 | 1,430.5 | 1,178.0 |
| 100 m² (1,076 ft²) | 670.9 | 573.6 | 513.2 | 422.6 |
Each cell is area × 0.6233 × Cr × 0.90. In metric the equivalent statement is simpler: 100 m² collects 100 litres per millimetre before losses, or 90 × Cr litres per millimetre after them.
How to turn the yield into a system design
Yield tells you the ceiling; storage and demand timing set the reality. A system's actual delivery is limited by whichever runs out first — rain, tank volume, or the demand's willingness to be served when water happens to be available. The standard design method is a monthly water balance: for each month, add that month's harvest, subtract that month's demand, carry the balance forward, and cap it at tank capacity. Any month that goes negative is a month you need mains make-up; any month capped at the tank ceiling is water you spilled.
Size the tank against your dry season, not your annual total. A climate with 40 inches spread evenly needs far less storage than one with 40 inches concentrated in four months. As a starting point, size for the volume of demand across the longest reliably dry stretch, then check that number against the cost of a tank that size.
Check the overflow against your biggest storm, not your average one. The table's largest rows exist for this. A 1,500 ft² roof under a 3 inch storm delivers over 2,100 gallons in a few hours; the overflow path, the gutters and the downpipes all have to pass that flow safely to somewhere that will not undermine a foundation. Where a design storm figure is needed for the site's overall drainage, the SCS curve number runoff calculator is the standard method.
Do not expect the water bill to justify the system. At typical US water rates, saving $167 a year against a tank, pump, filters and installation that may cost several thousand dollars is a payback measured in decades. Rainwater systems are usually built for drought resilience, stormwater credit, irrigation quality, code compliance in water-stressed jurisdictions, or a green building rating — the bill saving is a bonus, not the case.
Rainwater is not potable water
Roof runoff carries bacteria, roof material leachates, atmospheric deposition and whatever landed on the surface between storms. Treat it as non-potable unless it is filtered, disinfected and permitted as potable under your local code. ARCSA/ASPE/ANSI 63 and most adopted plumbing codes require any non-potable distribution to be clearly labelled, physically separated from potable piping, and protected against backflow, with an air gap on any municipal make-up connection. Several US states also regulate or restrict harvesting itself under water-rights law — check before you install, not after.
Mistakes that inflate a yield estimate
- Using sloped roof area. Rain falls vertically. Use the plan footprint, or you will overstate a steep roof by 20 to 40 percent.
- Ignoring tank overflow. This calculator gives interception, not delivery. Without storage sized to the demand pattern, a large share of the annual harvest spills.
- Using a single wet year's rainfall. Design against a long-term normal, and stress-test against a dry year at perhaps 70 percent of it.
- Applying first flush once a year. The diverter dumps on every storm. Multiply by event count.
- Assuming a runoff coefficient of 1.0. No real roof achieves it. Even metal loses water to wetting and evaporation.
- Forgetting winter. In freezing climates the system is drained and out of service for months, and that unavailable period is not reflected in an annual rainfall total.
- Counting sewer savings you do not get. Many utilities bill outdoor irrigation without the sewer component, so the rate you avoid may be lower than the combined rate on your bill.
Related calculations and where this method stops
Rainwater harvesting sits at the intersection of two disciplines that use the same rainfall data for opposite purposes. A harvesting designer wants to know how much water a surface delivers; a stormwater engineer wants to know how much it sends downstream, which is the same physics with the sign reversed. The curve number method in the TR-55 runoff calculator is the standard tool for the second, and in many jurisdictions a cistern earns credit against a stormwater requirement precisely because it retains volume that would otherwise leave the site.
What this calculator does not model is time. It is a volume balance over whatever period you enter, and it says nothing about whether the water is there on the day you want it. For a real design, take the yield-per-inch figure it gives you, apply it to a month-by-month rainfall record, and run the storage balance. That is the only way to size a cistern honestly.
It also stops at the tank. Pump energy, ultraviolet or filtration equipment, and any heating of the harvested water all carry an ongoing energy cost — one that can be quantified alongside your other household energy through the water heating cost calculator and rolled into a household carbon footprint. Municipal water also carries an embodied energy cost for treatment and pumping that harvesting avoids, which is one of the more defensible non-financial arguments for these systems.