Planning guide
What size rainwater tank do I need?
Compare the rain your roof can collect with the water you intend to use, then test several tank capacities against the daily rainfall record. Annual rainfall estimates potential collection; it does not tell you how much storage will bridge a dry spell.
Quick answer: there is no universal size for a roof or household. Compare the connected roof, daily rainfall and intended use across several tank sizes. A larger tank helps only when it can retain water that would otherwise spill before a later dry day.
Try the garden tank-size example → or enter your own location and inputs.
Start with roof yield, water use and rainfall timing
A tank size needs four inputs: connected roof area, rainfall, collection efficiency and demand. Use the roof-yield formula to estimate incoming water and the water-demand guide to estimate outgoing water. Then compare candidate capacities in the free rainwater tank calculator. Check both demand coverage and the longest shortfall; a high average coverage can still hide days without enough water.
There is no universal correct size per household or square metre of roof. The reproducible examples below show why the same annual rainfall can favour very different storage volumes.
1. Define the job before the tank
Start with the water use you want the system to support. “Collect as much rain as possible” is not yet a design goal. Garden irrigation, toilet flushing, wash-down, and emergency reserve have different timing, quality, plumbing, and regulatory requirements. This planner uses a constant daily non-potable demand during the months you select, with zero demand in other months.
Write down what counts as success. It might be reducing mains use, supplying a stated daily volume, making it through a typical dry spell, or avoiding an oversized tank with little extra modeled benefit. The metric you care about changes how you read the result.
2. Measure connected roof footprint
Use the horizontal plan area that actually drains through connected gutters to the proposed tank. Do not use the sloped roof surface, and do not include roof sections that discharge elsewhere. Split roofs, valleys, blocked gutters, and downpipe routing can make the connected area smaller than the building footprint.
In metric units, the catchment identity is simple:
1 mm of rain × 1 m² of connected roof = 1 litre before losses
For example, 10 mm falling on 120 m² produces 1,200 litres before losses. At 85% collection efficiency, the modeled inflow is 1,020 litres. That one-day number does not determine tank size because yesterday’s storage, today’s demand, overflow, and the next dry period still matter.
3. Choose daily rainfall evidence
A yearly total discards sequence. Two locations with the same annual rainfall can behave differently when one receives frequent modest rain and the other receives a few intense storms separated by long dry periods. Use multiple completed years of daily data so the simulation includes a range of sequences.
The planner’s default NASA POWER source gives consistent global coverage, but it is modeled grid data rather than a gauge on your property. A nearby quality-controlled gauge may be more representative. When the purchase depends on small differences, compare both sources and investigate why they disagree. The rainfall data guide explains the checks.
4. Use an efficiency range, not a magic percentage
Not every litre reaching a roof reaches useful storage. First-flush diversion, filters, gutter overflow, wind, splash, leakage, maintenance, and unusable tank volume all reduce yield. The U.S. Department of Energy review cites a common collection-efficiency range of 75%–90%, but a generic range cannot describe every roof and system.
Run at least a conservative and an optimistic value. If the preferred capacity changes sharply across that range, inspect the physical design before trusting the central case.
5. Estimate demand honestly
Demand is often less certain than tank volume. List each intended use, estimate volume per event, and multiply by frequency. Convert the result to an average daily value for the months when water is used. Then run low, central, and high cases. For seasonal irrigation, choose the active months in the calculator; rainfall still fills the tank outside the season. The model does not represent different amounts within an active month or weekday-only use.
6. Compare a useful capacity ladder
Enter actual sizes you could buy or fit, not dozens of tiny increments. Include one clearly small tank, several realistic choices, and one large case. The large case shows the best coverage available among the tested options; it is not automatically practical or economical.
The highlighted “near-plateau tank” is the smallest entered capacity within two percentage points of the best modeled coverage. That rule exposes diminishing returns among your choices. It does not include tank price, excavation, structural support, pumps, treatment, maintenance, local approval, or the value of a reserve.
7. Read all outcomes together
- Demand covered measures the share of requested litres supplied. It can be high even when some individual days fail.
- Full-demand days shows how often the entire daily draw was available.
- Water supplied per year helps compare modeled mains-water displacement, before any financial assumptions.
- Overflow shows captured inflow that could not fit. More capacity can reduce it, but overflow is not automatically wasted if safe drainage is valuable.
- Longest shortfall exposes dry sequences that an annual percentage can hide.
8. Stress-test the decision
The calculator now checks 25% lower, entered and 25% higher water use automatically. Read the sensitivity table to see how the same tank's coverage and longest shortfall change, alongside the size highlighted in each case. It keeps your selected months and other assumptions fixed. If you start with a quick roof-area estimate, try another area separately too; the water-use test does not resolve roof uncertainty.
Change one uncertain input at a time: rainfall source, demand, efficiency, starting level, and candidate sizes. A robust choice should remain understandable across plausible cases. If a tiny input change produces a large decision change, the correct conclusion is uncertainty—not extra decimal places.
Worked example: the same 600 mm, two very different storage results
This experiment isolates rainfall timing. We constructed two complete daily CSV files for 1 January–31 December 2025. They are synthetic teaching data, not NASA records, station observations, a climate comparison, or a forecast. Both total 600 mm, with no missing dates.
- Frequent rain: one event every five days, starting 1 January. There are 73 events of 600 ÷ 73 mm (about 8.22 mm) and 292 zero-rain days.
- Monthly storms: 50 mm on the first day of each month and zero on every other day. There are 12 wet days and 353 zero-rain days.
For both, use a 100 m² connected roof, 80% collection efficiency, 100 L/day demand, and an initially empty tank. Compare 500, 1,000, 2,000, 3,000, and 5,000 L. The year's potential harvest is 600 × 100 × 0.80 = 48,000 L; demand is 365 × 100 = 36,500 L. An annual balance therefore looks sufficient in both cases. It cannot tell us how much of that water fits in the tank when it arrives.
| Tank | Frequent rain: demand covered | Monthly storms: demand covered | Monthly storms: longest shortfall |
|---|---|---|---|
| 500 L | 100% | 16.4% | 26 days |
| 1,000 L | 100% | 32.9% | 21 days |
| 2,000 L | 100% | 65.8% | 11 days |
| 3,000 L | 100% | 98.1% | 1 day |
| 5,000 L | 100% | 100% | 0 days |
Why the 500 L tank behaves differently
A frequent-rain event produces about 657.5 L of inflow. The empty 500 L tank retains 500 L and spills about 157.5 L before the day's demand. That leaves 400 L after the first 100 L draw, enough for the following four dry days. The next event arrives on day six, so every day's demand is met in this constructed sequence.
A monthly 50 mm storm produces 4,000 L. An empty 500 L tank retains only 500 L and spills 3,500 L before use. It supplies five days at 100 L/day, then runs out until the next month. Twelve events supply only 6,000 of the 36,500 L demanded: 16.4%, despite the same total rainfall and potential harvest.
A 3,000 L tank covers 30 daily draws after each monthly storm. It falls short on the last day of each of the seven 31-day months, supplying 35,800 L altogether. Coverage is 98.1%, but full supply still fails on seven days. A 5,000 L tank covers all days here and finishes with 1,900 L; that remaining water is stored, not counted as supplied.
The highlighted tank does not mean zero shortfalls
The planner highlights 500 L for frequent rain and 3,000 L for monthly storms. The latter is within two percentage points of the best tested coverage (100%), so it qualifies under the near-plateau rule even with seven shortfall days. If uninterrupted supply is your goal, inspect the reliability and shortfall columns rather than relying on the highlighted capacity alone. None of these capacities is a purchase recommendation for an actual property.
Reproduce it, then stress-test it
- Open the calculator and choose either example button. The CSV and the inputs above load together. Press Compare tank sizes.
- For a separate check, download frequent-rain daily data or monthly-storm daily data, choose My daily CSV, and enter the stated inputs.
- Compare against the table and the complete modeled results CSV. The CSV includes supplied water, overflow, ending storage, and shortfalls.
- Raise demand to 150 L/day or reduce collection efficiency to 60%. Change one input at a time and inspect what becomes limiting. Replace the example with a long, checked local daily record before making a real planning decision.
A small display difference is intentional: this example reports a calendar-year total of 600 mm across 365 days. The calculator annualizes rates using 365.2425 days, so its average annual rainfall shows 600.4 mm. Its annualized supplied and overflow litres also differ slightly from the exact one-year totals explained above. Demand coverage uses the actual modeled totals.
Three practical scenarios to explore
These examples combine an explicit use estimate with the synthetic 2025 rainfall files above. Each file totals 600 mm over 365 days. They illustrate the calculation; they are not real properties, observed local weather, or recommended watering rates.
All cases start empty, collect 80% of roof rainfall, and test 500, 1,000, 2,000, 3,000 and 5,000 L. The gardening cases use daily averages only in their selected months. Choose your own season for your climate.
Small garden with frequent rain
An 80 m² connected roof supplies a 35 m² garden. A planned 10 mm per week gives 35 × 10 ÷ 7 = 50 L/day, used from May through September (153 days). With frequent-rain data, every tested tank supplies 100% of that modeled demand. The highlighted 500 L size is enough for this deliberately regular pattern; a different dry spell could change the result.
Load the small-garden scenario →
Larger garden with monthly storms
A 120 m² roof supplies a 70 m² garden. At the illustrative 10 mm per week, demand is 100 L/day from April through October (214 days). With monthly-storm data, a 500 L tank supplies 16.4% of demand, 3,000 L supplies 98.1%, and 5,000 L supplies 100%. The highlighted 3,000 L tank still has four days with unmet demand, with a longest shortfall of one day. The storage comparison matters even though the yearly rainfall is unchanged.
Load the larger-garden scenario →
Year-round toilet flushing
A 100 m² roof supplies two people, each using five flushes per day at an assumed 6 L per flush: 2 × 5 × 6 = 60 L/day throughout the year. Check the volume for your actual fixtures and whether this use is locally permitted. In the monthly-storm pattern, 500 L supplies 27.4% of demand and 2,000 L supplies 100%. The model highlights 2,000 L for this constructed case.
Load the toilet-flushing scenario →
| Tank | Small garden | Larger garden | Toilet flushing |
|---|---|---|---|
| 500 L | 100% | 16.4% | 27.4% |
| 1,000 L | 100% | 32.7% | 54.8% |
| 2,000 L | 100% | 65.4% | 100% |
| 3,000 L | 100% | 98.1% | 100% |
| 5,000 L | 100% | 100% | 100% |
Load a case, press Compare tank sizes, and check the table against the result. Change one assumption at a time, then use historical NASA rainfall or a checked local CSV for your own location. Full-demand day percentages count only days with positive demand, so winter months do not artificially improve the gardening cases' reliability.
Is a 5,000 litre tank big enough?
In the larger-garden example, a 120 m² roof supplies 100 L/day from April through October using an illustrative 600 mm synthetic rainfall pattern. A 3,000 L tank covers 98.1% of modeled demand and 5,000 L covers 100%. With different rainfall timing, demand or losses, both results can change. Load the 5,000 litre comparison example, then replace its assumptions with evidence for your own site.
In US units, 5,000 L is approximately 1,321 US gallons; a 5,000 US gallon tank is about 18,927 L. Check which unit a supplier uses and recalculate its nominal size. Before purchase, use the rainwater tank buying checklist to check access, equipment, delivery and installation requirements.
Already own a tank or barrel? Use the rain barrel watering calculator to check how long the water currently stored lasts, including a reserve and a higher-use case. Its optional gravity check compares water-surface height with an equipment-label minimum.