The runoff coefficient is the fraction of rain that a roof actually delivers to the tank: 0.9 for metal, 0.8 for tile and 0.5 for thatch, which absorbs much of the rain. A real system delivers less than the coefficient suggests because every rain event also loses the first-flush discard, so with 100 separate events a year at 1 mm each the usable share of 800 mm of rain falls from 0.9 to about 0.79.
Part of Cisterns and Rainwater Storage
The roof coefficient is the single multiplier that converts rain falling on a roof into water reaching the tank. It bundles the roof’s absorption, evaporation and splash losses into one number, and it is the least certain term in the yield formula.
The Coefficient Bundles Several Losses Into One Guess
The formula in the cisterns page is litres = area (m²) × rainfall (mm) × coefficient. Its values are:
| Surface | Coefficient |
|---|---|
| Metal roof | 0.9 |
| Tile roof | 0.8 |
| Thatch roof | 0.5 (absorbs a lot) |
| Concrete or plastic ground surface | 0.8 |
A metal roof’s 0.9 sits at the low end of the 90–95% efficiency given for metal and plastic roofs in Catchment Systems. What the coefficient allows for is water that wets the surface and never runs off, water evaporating before it reaches the gutter, water splashing past the gutter and water lost at leaky joints. Thatch absorbs the most and releases some of it slowly after the rain stops.
Discarded First Flush Comes Out Of The Yield On Top Of The Coefficient
The first-flush rule is to divert 1 L per m² of roof every time it rains (First Flush Diverter). That is 1 mm of rain on every event. The coefficient does not include it. The loss depends on how the rain falls, not on how much:
- 100 separate events a year × 1 mm = 100 mm lost out of 800 mm
- Usable rain = 800 − 100 = 700 mm
- Combined factor = 0.9 × (700 ÷ 800) = 0.9 × 0.875 = 0.79
The 100 events are a made-up number for illustration; count the events in your own rain log. In a climate of many light showers the discard is a large share, and for a 2 mm shower the diverter throws away half of it. In a climate of a few big storms it is small.
What You Need
- Your rain log — number of events and depth of each
- A tank with a straight-sided cross-section — so a dip-stick reading converts to litres
- A stick or ruler to read the level
- A clean roof — a test on a roof coated in dust or leaves gives a falsely low result
Method: Calibrate Your Own Roof
Step 1: Note the tank level before a rain event, with enough empty space that the tank cannot overflow during it.
Step 2: Read the gauge for the rain depth, in millimetres, after the event.
Step 3: Read the tank level again. Convert the rise to litres using the tank’s area.
Step 4: Divide. Measured litres ÷ (roof area × rain depth) is the effective coefficient for that event, including losses. Repeat for several events and average, because small events are dominated by the first-flush loss.
Worked Example
A 1.6 m diameter tank (the size of the ferrocement tank on the cisterns page) has a floor area of π × 0.8² = 2.01 m². A 40 m² metal roof receives 10 mm of rain.
- Potential: 40 × 10 = 400 L
- Predicted with 0.9: 400 × 0.9 = 360 L
- Predicted tank rise: 360 ÷ 2.01 = 179 mm
If the tank rises 120 mm, the delivery was 120 × 2.01 = 241 L. Effective coefficient = 241 ÷ 400 = 0.60. The first-flush discard of 1 L/m² on this roof is 40 L, which accounts for 0.1 of the 0.3 shortfall (40 ÷ 400); the rest is gutter spill, leaks or a dirty roof. This tells you where to look.
Common Mistakes
- Applying the 0.9 to a leaking gutter system. The coefficient describes the roof; a sagging gutter loses water the coefficient never counted.
- Quoting an annual figure without a margin. The coefficient varies with rain intensity and roof age. Do not read a precision into it that the data lacks.
- Treating thatch like tile. At 0.5, the same area yields five-ninths of what metal does (0.5 ÷ 0.9), and with poorer water quality.
What’s Next
- Learn next: Dry Spell Bridging — turning the yield estimate into a tank size
- Combine with: Leaf Screens — a blocked screen is a hidden loss that no coefficient covers