Numbered List · Technical Perspective · Published: October 14, 2026
Table of Contents
Key Takeaways
- The whole yield model rests on one identity: 1 mm of rain on 1 m² of roof is exactly 1 litre, so an 800 mm rainfall year on a 100 m² roof is 80,000 litres of gross potential before losses.
- Real harvest is that number multiplied by a runoff coefficient: metal roofs sit near 0.90-0.98, membrane and shingle roofs near 0.80-0.90, and green roofs can fall to 0.20-0.35 [4].
- First-flush diversion is the cheapest quality upgrade in the system: a minimum design criterion is diverting the first 0.5 mm of rainfall, with 0.5 L/m² for low-pollution roofs and up to 2 L/m² for high-pollution sites [2][3].
- Stored rooftop rainwater is not automatically clean: in a Philadelphia barrel study, lead was detected in 88.5% of barrels (3-1,282 µg/L) and zinc in 88.5% (29-2,660 µg/L) [5].
- Microbiology drives the disinfection decision: E. coli turned up in 16 of the 26 barrels tested in the same study’s summer round (62%), with 19% of samples exceeding the 235 CFU/100 mL single-sample benchmark [5].
- A practical storage rule of thumb from the Texas Manual: a cistern holding roughly a three-month water supply is adequate for most household use, with purchased water available to bridge longer dry spells [6].
Number 1: The Yield Identity — 1 mm × 1 m² = 1 L
Everything in rainwater harvesting starts from a piece of arithmetic that feels too simple to be load-bearing: one millimetre of rain falling on one square metre of catchment is one litre of water. A millimetre is a thousandth of a metre of depth; spread over a square metre it is a thousandth of a cubic metre, which is a litre. No coefficients, no safety factors, no jurisdictional quirks — it is geometry.
The identity matters because it converts the two things an owner usually knows — roof size and local rainfall — directly into litres, the currency of every downstream decision. A 200 m² warehouse roof under 900 mm of annual rain is a 180,000-litre gross opportunity. Whether any of it becomes usable water depends on the five numbers that follow, but every sizing conversation should begin by stating the gross number out loud, because it caps everything else in the system.
Number 2: The Runoff Coefficient — 0.75 to 0.98
The first haircut comes from the roof itself. Not every millimetre that lands on the catchment reaches the tank: some wets the surface, some evaporates, some splashes off at gutters, and some sits in the texture of the material. The runoff coefficient expresses the share that survives. Published design tables put smooth metal roofs at 0.90-0.98, flat membrane roofs at 0.85-0.95, asphalt shingles at 0.80-0.90, clay or concrete tile at 0.75-0.85, and green roofs — after their own retention layer has had its share — at 0.20-0.50 [4]. The same tables note the values trace to ARCSA/ASPE 63 and the Texas Water Development Board manual, which is why the numbers repeat across regional guidance.
The design implication is bidirectional. Overestimate the coefficient and the tank under-delivers through dry seasons; underestimate it and the owner pays for storage that never fills. Prudent practice applies the coefficient honestly, then uses a further safety factor for system losses — gutter overflow, first-flush diversion, filter losses — rather than inflating the coefficient itself. On a 150 m² tile roof with coefficient 0.80 under 800 mm of rain, the honest starting estimate is 96,000 litres per year, not the 120,000 the identity alone promises.
Number 3: First Flush — 0.5 mm as the Floor
The first rain after a dry spell rinses the roof’s accumulated cargo — dust, pollen, leaf debris, insect bodies, bird and rodent droppings — into the tank unless the system deliberately diverts it. The SOPAC guidance sets the minimum design criterion at diverting the first 0.5 mm of rainfall, with the diversion volume computed as roof length × width × 0.5 [2]: a 10 m × 10 m roof therefore needs a device that holds at least 50 litres before clean water is allowed through.
Practice ranges above that floor with contamination. Field guidance for rooftop systems puts low-pollution sites at about 0.5 litres per square metre of roof and high-pollution environments — heavy tree cover, bird roosting, nearby traffic — at up to 2 litres per square metre [3]. The Texas Manual frames the same question in imperial units: one rule of thumb diverts a minimum of 10 gallons per 1,000 square feet of collection surface, while an Australian vendor study recommended between 13 and 49 gallons per 1,000 square feet, precisely because the right answer depends on dry days, debris load and roof type [1]. The engineering lesson is that first-flush sizing is a calibration, not a constant: it should be revisited when roof surroundings change, and the diverter must drain between events so it is empty when the next storm arrives.
Number 4: Lead in Storage — Detected in 88.5% of Barrels
Collected rainwater inherits the roof’s material history. The most instructive recent dataset comes from a Greater Philadelphia study of roof-harvested rainwater barrels — 38 samples from 34 barrels — analyzed for metals and microbial indicators alongside qPCR pathogen screening [5]. Lead was detected in 88.5% of the barrels at concentrations from 3 to 1,282 µg/L among positive samples. Zinc, the signature of galvanized roofing and metal fixtures, appeared in the same 88.5% of barrels at 29-2,660 µg/L [5].
Two things follow. First, roof and plumbing materials are water quality devices: galvanized steel sheds zinc, older flashings and solders shed lead, and the barrel concentrates what they shed. Second, “clean until proven dirty” is the wrong default for stored roof water. The Philadelphia study is not an outlier argument — it is a reminder that quality varies by roof type, season and first-flush discipline, and that a system intended for irrigation tolerates numbers that a potable system never would [5].
Number 5: Microbiology — 62% Detection, 19% Over the Line
The same barrel study found E. coli in 16 of the 26 barrels tested in its summer round — 62% — with five samples, or 19%, exceeding the 235 CFU/100 mL single-sample recreational water benchmark, and Enterococcus present in 20 of the 26 (77%) [5]. Molecular screening extended the picture: Legionella species DNA was detected in 57.9% of samples, though the pathogenic species L. pneumophila was not found; opportunistic organisms such as non-tuberculous mycobacteria and Acanthamoeba also turned up [5]. Birds and mammals on the catchment are the usual reservoir; warm storage does the rest.
This is the number that decides the disinfection question. For toilet flushing and irrigation, microbial load is a managed nuisance. For anything involving human contact or drinking, storage alone is not treatment: household guidance for harvested rainwater converges on a barrier of chlorination, ultraviolet disinfection, boiling, or certified filtration applied after storage, combined with sanitary inspection of the catchment itself. The monitoring consequence is practical — residual chlorine belongs on the instrument list for any chlorinated rainwater loop, and a tank with E. coli history needs verification sampling, not optimism.
Number 6: Storage Sizing — The Three-Month Yardstick
Storage is where hydrology meets budget, and it is the number owners most often get wrong in both directions. A tank that is too small fills and overflows in the first storm of the season, wasting exactly the water the system was built to capture; a tank that is too large amortizes slowly and sits half-empty through marginal months. The Texas Manual’s working rule of thumb for household systems is that a cistern containing roughly a three-month supply is adequate for most home use, with the practical caveat that water can be purchased to top up during longer dry periods [6].
The sizing arithmetic follows directly from the earlier numbers. Daily demand × days of autonomy gives the required volume; the yield identity with the runoff coefficient says how fast refills arrive; and the first-flush volume subtracts from every event. A household drawing 150 L/day wants about 13,500 litres for a 90-day autonomy target — but if the catchment only yields 96,000 litres a year, a 13,500-litre tank will refill slowly, and the honest design either trims demand, extends the catchment, or accepts purchased top-ups. Sizing storage against demand and yield together is what separates a working system from a decorative one.
The Design Table
| # | Number | Typical value and source | Design consequence |
|---|---|---|---|
| 1 | Yield identity | 1 mm on 1 m² = 1 L | Converts roof size and rainfall straight into litres |
| 2 | Runoff coefficient | 0.90-0.98 metal; 0.75-0.85 tile; 0.80-0.90 shingle; 0.20-0.50 green [4] | First haircut on gross yield; material choice is a water choice |
| 3 | First flush | 0.5 mm minimum [2]; 0.5 L/m² low-pollution, up to 2 L/m² high-pollution [3] | Sets diverter volume; the main quality barrier before storage |
| 4 | Lead in barrels | Detected in 88.5% of barrels; 3-1,282 µg/L in positives [5] | Roof materials are part of water quality; test before potable use |
| 5 | Microbiology | E. coli in 62% of barrels; 19% over 235 CFU/100 mL [5] | Decides disinfection barriers for contact or potable reuse |
| 6 | Storage | About a three-month supply for most household use [6] | Balance demand × autonomy against refill rate and budget |
Where Instrumentation Fits
Rainwater systems drift quietly: catchment chemistry changes with roof work, first-flush diverters stick, and disinfection lapses invisibly. Continuous conductivity tracks salt-level changes and first-flush pass-through; residual chlorine verifies that any chlorinated loop remains protected; pH and turbidity flag catchment disturbance after storms. Shanghai ChiMay’s approach for decentralized and non-potable water systems runs through digital multi-parameter sensing — four parameters from one probe — paired with the simple conviction that sensors should feed management data, not just dashboards. For a system whose six numbers are already known at design time, continuous measurement is what confirms they still hold in operation.
The Six-Number Summary
Harvest forecasting, quality management and storage sizing in rainwater systems reduce to six numbers: the yield identity, the runoff coefficient, the first-flush volume, the lead and zinc legacy of the roof, the microbiology that sets the disinfection requirement, and the storage autonomy the owner actually needs. Designers who write all six down before buying a single component consistently build systems that survive their first dry season; the numbers are stable, published and easy to defend to any reviewer who asks.
References
- Texas Water Development Board — The Texas Manual on Rainwater Harvesting: first-flush rules of thumb of 10 gallons per 1,000 sq ft and the Australian vendor recommendation of 13-49 gallons per 1,000 sq ft. http://www.twdb.texas.gov/publications/reports/contracted_reports/doc/2003483510_RainwaterHarvesting.pdf
- SOPAC (Mosley, 2005) — Water Quality of Rainwater Harvesting Systems: minimum first-flush design criterion of diverting the first 0.5 mm of rainfall, with the length × width × 0.5 calculation. https://islaurbana.org/wp-content/uploads/2022/10/pluvioteca-water-quality-of-rainwater-harvesting-systems.pdf
- Agriculture.Institute — A Comprehensive Guide to Rooftop Rainwater Harvesting: first-flush guidance of about 0.5 L/m² for low-pollution roofs and up to 2 L/m² for high-pollution environments. https://agriculture.institute/soil-water-conservation/comprehensive-guide-rooftop-rainwater-harvesting
- DrainageCalculators — Rainwater Cistern Sizing Calculator: runoff coefficient table (min/typical/max) by collection surface, sourced to ARCSA/ASPE 63 and the TWDB rainwater harvesting manual. https://drainagecalculators.com/calculators/tier-2/cistern-sizing
- MDPI, Water (2018) — Assessment of Water Quality in Roof-Harvested Rainwater Barrels in Greater Philadelphia: lead detected in 88.5% of barrels (3-1,282 µg/L in positive samples), zinc in 88.5% (29-2,660 µg/L), E. coli in 16 of 26 barrels with 19% of samples above 235 CFU/100 mL, Enterococcus in 20 of 26, and qPCR detection of Legionella spp. in 57.9% of samples. https://www.mdpi.com/2073-4441/10/2/92/html
- Watershed Committee of the Ozarks — Rainwater Harvesting Manual for Southwest Missouri: relays the Texas Manual guidance that a cistern holding about a three-month supply is adequate for most household use. https://watershedcommittee.org/wp-content/uploads/2013/06/Rainwater-Manual.pdf
About the author: Written by the Shanghai ChiMay Technical Editorial Team — engineers specializing in multi-parameter water quality sensing for decentralized, rain-fed and non-potable reuse water systems.
