A butterfly roof puts every drop of water in one place, and that place is indoors.
Mixed-slope · single-family, small commercial, and additions
Two planes falling inward to a central valley. The form is deliberate and it does real work — high daylight, a collectable water supply, a distinctive profile. It also removes the escape route every other roof shape relies on.
What is a butterfly roof, and what is the catch?
A butterfly roof is two planes sloping inward to a central valley, so every drop of water on the roof arrives at one interior low point instead of running off an edge. That point sits over occupied space. It needs a waterproofing membrane, a drain sized for the whole roof, and a second exit that works when the first one blocks.
The short versionSection link
Six things to establish about any butterfly roof before anything else is decided. The third one is why this page exists.
- The shape
- Two planes falling inward to a central valleyAlso called an inverted gable or a V-roof. The planes need not be equal, and on many real examples they are not.
- Where the water goes
- One interior low point, over occupied spaceEvery other common shape sheds over an edge, where a failed gutter spills onto the ground outside the wall.
- The valley’s own slope
- Zero, unless somebody designed some inTwo planes falling towards each other meet along a level line. The fall along the trough is a separate design decision, not a by-product of the roof pitch.
- The covering in the valley
- Normally a membrane, not a shingle or a tileWater-shedding coverings need slope and lap. A trough that can stand water is a waterproofing problem, and the adopted code sets minimum design slopes covering by covering.
- The second exit
- A requirement in the adopted codes read for this page, not an upgradeWhere an upstand traps water if the primary drain blocks, a secondary overflow drain or scupper is required. Which edition applies to you is your jurisdiction’s answer, not this page’s.
- The maintenance obligation
- Continuous, and not cosmeticLeaves in a gable’s gutter are untidy. Leaves in a butterfly’s valley are the blockage the whole assembly is designed around.
This page treats the valley as a low-slope waterproofing problem. Here is when that framing is wrong.Section link
The position taken here is that a butterfly roof is a legitimate form whose central valley must be designed as a drained, membrane-waterproofed trough with a second exit — not as a big version of an ordinary roof valley. That position has real limits, and several of them come from the same codes this page quotes.
Best when
- The building genuinely wants what the shape gives: high side walls that can carry glazing above head height, a low centre that reduces apparent bulk, and a single collection point if rainwater is being harvested.
- The trough has been given its own longitudinal fall by design — warped planes, a tapered build-up, or a purpose-framed gutter — rather than being left level and hoped over.
- The drainage is drawn and sized by someone who does drainage: the low point, the sump, the drain, the leader, and the overflow, all as one system rather than as a detail added at the end.
- The valley is reachable and maintainable. Somebody can get to it safely, on a schedule, with a way to see the drain and clear it.
- The interior below the valley can survive being wet once. Not because it should be, but because a roof whose failure mode routes water into a single room is a different risk over a garage than over a library.
- You are already committed to a membrane roof. If the building is going to have a low-slope system anyway, the butterfly geometry adds design work rather than a whole new trade.
Think twice if
- The trough discharges over an open edge with nothing above it. Then this page is over-engineering your roof: the adopted code quoted below requires drains at each low point only where roofs are not sloped to drain over roof edges, and the secondary-drainage requirement is written for roofs where an upstand entraps water. An open-ended trough is a different, and simpler, problem.
- It is an existing roof being re-covered. The adopted residential code read for this page carries explicit reroofing exceptions: for roofs that provide positive drainage, replacement is not required to meet the minimum design slope, and secondary overflow drains or scuppers are not required to be added. “You must add an overflow” is not automatically true on a re-roof.
- You are in deep-snow country and the answer being proposed is simply a bigger drain. FEMA names valleys, sawtooth roofs, and parapeted flat roofs as geometries that collect drifted snow, and blocked or badly maintained drainage as a route to concentrated melt loads. In that climate the trough is a snow problem before it is a rain problem, and it is a design question rather than a product question.
- The motive is only the look, and the budget cannot carry the obligation. A butterfly roof that nobody maintains is worse than an ordinary roof that nobody maintains, because the failure is concentrated instead of distributed.
- The building is historic or the roof is a character-defining feature. Then the shape is not negotiable and the conversation is about how to make it work, which is a different conversation from whether to build one.
- The roof predates about 1990 and the work would disturb old built-up felts, mastics, or flashing cements. Those may contain asbestos, and testing comes before disturbance.
What changes the answer
- Whether the trough is enclosed by upstands, parapets, or walls, or is open at one or both ends. That single fact changes which code provisions apply.
- The design rainfall intensity at the address, which comes from the precipitation-frequency data the adopted plumbing code points to — not from any figure on this page.
- The plan area feeding the trough. Runoff volume follows the roof’s footprint, so a wide butterfly concentrates far more than a narrow one at the same pitch.
- The slope of the two planes, which decides how fast water arrives at the trough but nothing at all about how it leaves.
- The climate: freeze-thaw and snow change the trough from a hydraulic problem into a load and ice problem.
- The deck and framing. What a trough may hold before it matters is a structural determination for a licensed design professional looking at this building.
- What your jurisdiction has actually adopted and amended, and what its authority having jurisdiction will require on a permit.
- Whether the water is being collected on purpose. A harvesting system changes the trough from a nuisance into an asset, and adds its own filtration and overflow obligations.
Every other roof shape has an escape route. This one gave it up on purpose.Section link
A gable, a hip, and a shed all end at an edge with air under it. A butterfly ends in the middle, over a room. Everything difficult about the shape follows from that single decision.
- The two planes. They fall inward instead of outward. No eave on the building sheds roof water, so no gutter on the outside of the building is doing the job a gutter normally does.
- The central valley. The valley where the two planes meet, and the only low point on the roof. It is usually drawn not as a knife edge but as a flat trough a few feet wide, because a knife edge is impossible to flash and impossible to walk.
- Drainage concentration. Every arrow on both planes ends in the same place. A gable divides its catchment between two eaves and spreads each half along the whole length of a gutter. A butterfly does the opposite: it collects the entire roof and presents it to one outlet.
- Occupied space. The low point is inboard of both walls, which means it is over a room. Water that fails to leave does not spill onto the garden. It stands, and then it finds a way in, and the way in is over something.
- The membrane. A single waterproofing sheet through the trough, turned up both planes and terminated well above any level the water can reach. This is the detail that separates a butterfly roof that works from one that leaks early, and it is the reason the trough is normally a different system from the field of the roof.
- The fall the design has to add. In the lower panel the dashed line is what the two planes give the trough on their own: a level line. The real trough falls away below it towards the drain. That wedge is slope somebody put there on purpose, and if nobody did, there is none.
- The primary drain, in a sump. A drain bowl set in a shallow dished depression, so the water actually gets to it rather than standing in a ring around a clamping ring. NRCA “recommends drain sumps at all internal drain locations,” noting that sumps “reduce the potential for localized ponding that could be caused because of a buildup of materials and a drain clamping ring.”
- The overflow. A secondary drain or a scupper through the upstand, with its inlet set above the roof low point so it does nothing until the primary path stops working. The shaded pool between those two levels is the water the roof is expected to carry in the meantime. The arrow leaving it is the point: overflow drainage discharges where somebody notices.
- The raised outer walls. The high side of each plane sits on a tall wall, and a tall wall can carry glazing above head height. This is a real and permanent benefit of the geometry, and it is worth saying so on a page that spends most of its length on the drainage.
Why the concentration is the whole story
A roof does not decide how much rain falls on it; the sky does. What a roof shape decides is where that water is presented, in what quantity, and how many independent places it can leave from. A simple gable has two catchments, two eaves, and a continuous line of exit along each. If a section of gutter blocks, the water spills over its front edge and lands outside the building. That overflow is unplanned, it is ugly, and it works.
A butterfly has one catchment and one exit. There is no unplanned overflow, because there is no edge to spill over. This is why the secondary drainage requirement in the codes read for this page is not a nicety. On this shape it is the substitute for the behaviour every other roof gets for free.
What “premium detailing” actually means here
The phrase gets used about butterfly roofs as though it meant expensive materials. It does not. It means a short list of specific things that a conventional roof either does not need or gets away with doing badly: a trough with its own designed fall; a waterproofing membrane rather than a water-shedding covering at the low point; a drain in a sump rather than a drain in a flat field; a leader that goes somewhere; a second exit at a higher level; and terminations high enough that a temporary pond does not reach them. Each is cheap in isolation. Missing any one of them is expensive.
Every valley is shallower than the roof that feeds it. A butterfly’s is level.Section link
This is the fact the whole shape turns on, and it is derivable in one line. NRCA publishes the ordinary case; the butterfly is the same rule taken to its limit.
NRCA states the general principle: “for steep-slope roofs where two roof areas of equal slope intersect to form a valley, the resulting valley slope is less than that of the two adjacent roof surfaces,” and gives the number — two 4:12 planes make a valley of “only about 3:12.”
The reason is geometry, not roofing. Water on each plane runs straight down its own fall line. The valley has to run along the bisector between those two fall directions, and travelling along a bisector means you make less downhill progress per foot of travel than you would going straight down either plane. If the two fall directions are separated by an angle of 2θ in plan, then:
valley slope = plane slope × cos θ
Two roofs meeting at a right angle — the ordinary cross-gable valley — have fall directions 90° apart, so θ is 45° and the factor is cos 45° = 0.71. A 4:12 plane gives a 2.83:12 valley, which is NRCA’s “about 3:12.” That agreement is the check on the formula.
Now push it. On a butterfly roof the two planes face each other directly: their fall directions are 180° apart, θ is 90°, and cos 90° is zero. Whatever the pitch of the planes, the line where they meet is level.
| Angle between the two fall directions, in plan | Valley slope as a fraction of plane slope | Valley slope where both planes are 4:12 | What this shape is |
|---|---|---|---|
| 60° | cos 30° = 0.87 | about 3.5:12 | Two wings meeting at a shallow angle |
| 90° | cos 45° = 0.71 | about 2.8:12 — NRCA publishes “about 3:12” | The ordinary cross-gable valley |
| 120° | cos 60° = 0.50 | 2.0:12 | A shallow-angled V in plan |
| 150° | cos 75° = 0.26 | about 1.0:12 | Very nearly facing each other |
| 180° | cos 90° = 0.00 | Level — no fall at all | A butterfly roof |
Read this table one item at a time
60°
- Valley slope as a fraction of plane slope
- cos 30° = 0.87
- Valley slope where both planes are 4:12
- about 3.5:12
- What this shape is
- Two wings meeting at a shallow angle
90°
- Valley slope as a fraction of plane slope
- cos 45° = 0.71
- Valley slope where both planes are 4:12
- about 2.8:12 — NRCA publishes “about 3:12”
- What this shape is
- The ordinary cross-gable valley
120°
- Valley slope as a fraction of plane slope
- cos 60° = 0.50
- Valley slope where both planes are 4:12
- 2.0:12
- What this shape is
- A shallow-angled V in plan
150°
- Valley slope as a fraction of plane slope
- cos 75° = 0.26
- Valley slope where both planes are 4:12
- about 1.0:12
- What this shape is
- Very nearly facing each other
180°
- Valley slope as a fraction of plane slope
- cos 90° = 0.00
- Valley slope where both planes are 4:12
- Level — no fall at all
- What this shape is
- A butterfly roof
Original derivation, Understanding Roofing. The formula assumes two planes of equal slope and a straight intersection. Real butterfly roofs escape the bottom row only because a designer deliberately broke one of those assumptions — by warping the planes, by building fall into a tapered layer above the deck, or by framing a separate falling gutter along the trough. None of that happens by accident, and none of it is visible from the street. NRCA’s 4:12 figure is quoted from Professional Roofing, August 2018; the full citation is in the source list below.
This is why “what pitch are the planes?” is the wrong first question about a butterfly roof and “how much fall does the trough have?” is the right one. The planes could be 8:12 and the trough could still be dead level. They are independent numbers, and only one of them decides whether the roof drains.
How much water arrives at one point, and how fast it stacks upSection link
One hypothetical roof, carried through the arithmetic. The numbers are arithmetic about that roof, not a determination about yours, and the rainfall rate is a round number chosen to make the sum legible.
Start with the unit. One square foot of plan area under one inch of rain receives 144 cubic inches of water, and a US gallon is 231 cubic inches, so 144 ÷ 231 = 0.623 gallons. The Texas Water Development Board’s rainwater harvesting manual uses the same figure: “approximately 0.62 gallons per square foot of collection surface per inch of rainfall.”
Note plan area. Rain falls vertically, so the volume a roof collects follows its footprint, not its surface area. The same manual is explicit: “regardless of the pitch of the roof, the effective collection surface is the area covered by collection surface.” Pitch changes how fast the water gets to the trough. It does not change how much there is.
| Step | Working | Result |
|---|---|---|
| Plan area of the roof | 44 ft × 28 ft | 1,232 sq ft |
| Water per inch of rain on that footprint | 1,232 sq ft × 0.623 gal | 768 gallons |
| Assumed design rainfall intensity | A round number for arithmetic. The real figure for an address comes from precipitation-frequency data and the adopted plumbing code. | 3 inches per hour |
| Flow arriving at the trough | 768 gal × 3 ÷ 60 min | about 38 gallons per minute |
| The same footprint as a gable, per eave | 38 gpm split between two eaves, each 44 ft long | 19 gpm per eave, about 0.4 gpm per foot of gutter |
| Rate of rise in the trough if the outlet blocks | 38 gpm ÷ 7.48 = 5.1 cu ft/min, over a 3 ft × 44 ft trough (132 sq ft) | about 0.47 in per minute — an inch every two minutes |
| Weight of each inch of standing water in the trough | 62.4 lb/cu ft ÷ 12 = 5.2 lb per sq ft per inch, × 132 sq ft | about 690 lb per inch of depth |
| Depth at which the overflow in the quoted code begins to run | Inlet flow line 2 in above the low point of the roof served | about 1,400 lb standing, ignoring water on the planes |
Read this table one item at a time
Plan area of the roof
- Working
- 44 ft × 28 ft
- Result
- 1,232 sq ft
Water per inch of rain on that footprint
- Working
- 1,232 sq ft × 0.623 gal
- Result
- 768 gallons
Assumed design rainfall intensity
- Working
- A round number for arithmetic. The real figure for an address comes from precipitation-frequency data and the adopted plumbing code.
- Result
- 3 inches per hour
Flow arriving at the trough
- Working
- 768 gal × 3 ÷ 60 min
- Result
- about 38 gallons per minute
The same footprint as a gable, per eave
- Working
- 38 gpm split between two eaves, each 44 ft long
- Result
- 19 gpm per eave, about 0.4 gpm per foot of gutter
Rate of rise in the trough if the outlet blocks
- Working
- 38 gpm ÷ 7.48 = 5.1 cu ft/min, over a 3 ft × 44 ft trough (132 sq ft)
- Result
- about 0.47 in per minute — an inch every two minutes
Weight of each inch of standing water in the trough
- Working
- 62.4 lb/cu ft ÷ 12 = 5.2 lb per sq ft per inch, × 132 sq ft
- Result
- about 690 lb per inch of depth
Depth at which the overflow in the quoted code begins to run
- Working
- Inlet flow line 2 in above the low point of the roof served
- Result
- about 1,400 lb standing, ignoring water on the planes
Original worked example, Understanding Roofing. Every line is arithmetic and can be checked. The last two lines understate the real quantity, because once water stands in the trough it also runs a little way up both planes; the extra depends on the pitch and is not computed here. The 2-inch overflow height belongs to the City of Seattle’s adopted residential code and is quoted as an example of how such a provision is written — your jurisdiction’s figure is your jurisdiction’s to state. This is not a load determination, a drainage design, or a determination that any deck can carry any of it.
Compare the two middle lines. A gable of exactly the same footprint presents about four-tenths of a gallon a minute to each foot of gutter, along eighty-eight feet of edge, with the ground underneath. The butterfly presents thirty-eight gallons a minute to one hole, with a room underneath. The total is identical. The concentration is not, and concentration is the entire engineering problem.
The rate-of-rise line is the one worth remembering. An inch every two minutes means a blocked trough is not a slow-developing problem you notice next week. In a real storm it is a problem that develops during the storm, which is precisely why the second exit has to be built in rather than fetched when needed.
If you want to run this for a real roof, start with the anatomy and pitch pages, then use the calculators for area. But the drain, the leader, and the overflow are sized by somebody qualified to size them, against the code your jurisdiction adopted. This page does not do it and neither should a calculator.
Three ways water leaves a butterfly roof, and what each one fails fromSection link
The right answer depends on whether the trough is enclosed or open, on the climate, and on where the discharge can go. What does not vary is that a second path is needed wherever the first one can be trapped.
| Arrangement | How the water leaves | What it depends on | How it fails | Where the second path goes |
|---|---|---|---|---|
| Internal drain | Through a drain bowl in a sump at the trough low point, into a leader inside the building. | A designed longitudinal fall along the trough, a sump around the bowl, and a leader with somewhere to discharge. | Blockage at the bowl or in the leader. Freezing in the leader where it runs through unconditioned space. Localized ponding around the clamping ring where there is no sump. | A second, higher drain in the same sump or nearby, on its own leader. NRCA suggests 12 to 18 inches edge-to-edge between the two so both can be flashed properly. |
| Through-wall scupper | Sideways through an upstand or parapet at the low end of the trough, into a conductor head or straight to daylight. | The trough falling towards that wall, and an exterior face that can take the discharge. | Ice and debris at the opening, which is fully exposed to the weather. Staining and saturation of the wall below where there is no conductor head. | A higher overflow scupper in the same wall. Seattle’s building code requires that scuppers used for secondary drainage be sized to keep ponding within the design depth, with an opening dimension of at least 4 inches, and that the primary system’s flow not be counted when sizing them. |
| Open-ended trough | The trough simply runs out over an open edge — a canopy, a carport, a small addition — and the water falls off. | There being no upstand, parapet, or wall at that end able to trap water. | Nothing dramatic. It fails the way a gutter fails: it overflows in the wrong place and stains something. | None required by the provision quoted here, which is written for roofs where the perimeter construction extends above the roof such that water is entrapped. This is the one arrangement where the secondary-drainage argument does not apply. |
Read this table one item at a time
Internal drain
- How the water leaves
- Through a drain bowl in a sump at the trough low point, into a leader inside the building.
- What it depends on
- A designed longitudinal fall along the trough, a sump around the bowl, and a leader with somewhere to discharge.
- How it fails
- Blockage at the bowl or in the leader. Freezing in the leader where it runs through unconditioned space. Localized ponding around the clamping ring where there is no sump.
- Where the second path goes
- A second, higher drain in the same sump or nearby, on its own leader. NRCA suggests 12 to 18 inches edge-to-edge between the two so both can be flashed properly.
Through-wall scupper
- How the water leaves
- Sideways through an upstand or parapet at the low end of the trough, into a conductor head or straight to daylight.
- What it depends on
- The trough falling towards that wall, and an exterior face that can take the discharge.
- How it fails
- Ice and debris at the opening, which is fully exposed to the weather. Staining and saturation of the wall below where there is no conductor head.
- Where the second path goes
- A higher overflow scupper in the same wall. Seattle’s building code requires that scuppers used for secondary drainage be sized to keep ponding within the design depth, with an opening dimension of at least 4 inches, and that the primary system’s flow not be counted when sizing them.
Open-ended trough
- How the water leaves
- The trough simply runs out over an open edge — a canopy, a carport, a small addition — and the water falls off.
- What it depends on
- There being no upstand, parapet, or wall at that end able to trap water.
- How it fails
- Nothing dramatic. It fails the way a gutter fails: it overflows in the wrong place and stains something.
- Where the second path goes
- None required by the provision quoted here, which is written for roofs where the perimeter construction extends above the roof such that water is entrapped. This is the one arrangement where the secondary-drainage argument does not apply.
The third row is the honest exception to this page’s general position. The adopted residential code quoted here requires drains at each low point “unless roofs are sloped to drain over roof edges,” and requires secondary overflow drains or scuppers only where the roof perimeter construction extends above the roof so that water will be entrapped. A small butterfly whose trough daylights over an open edge is a genuinely simpler problem than the rest of this page describes. Confirm what applies where you live with your authority having jurisdiction.
Why the trough normally gets a membrane
Steep-slope coverings are water-shedding: overlapping units that pass water from one course to the next by gravity and lap. They need slope for that to work, and the adopted codes set the thresholds covering by covering. In the residential code read for this page, asphalt shingles may be used “only on roof slopes of 2 units vertical in 12 units horizontal…or greater,” and membrane systems — built-up, modified bitumen, thermoset and thermoplastic single-ply, sprayed foam and liquid-applied — carry a design slope requirement of not less than 1/4:12 “for drainage.”
Set those two facts against the geometry table above. The trough is the flattest part of the roof by construction, and it is the part most likely to hold water. A covering that relies on lap and gravity is being asked to work in the one place that supplies neither. That is why the trough on a competently built butterfly roof is normally a continuous waterproofing membrane, turned up both planes and terminated above the water line, regardless of what the planes above it are covered in.
It also means the transition — where the membrane stops and the plane covering starts — is a real detail with a real height, not a place where two trades meet and hope. Ask for that height in inches. Compare it against the depth at which your overflow would begin to run. If the second number is bigger than the first, the roof has a designed leak.
Which membrane is a separate question, answered by the same considerations as any other low-slope roof. See the membrane comparison and membranes at residential scale.
In a cold climate the trough is a snow problem before it is a rain problemSection link
Everything above assumes water that is trying to leave. Snow is not trying to leave, and ice is actively preventing the water behind it from leaving.
FEMA’s snow guidance is unusually direct about geometry. It identifies a valley — described as the “intersection at two opposite sloped roofs” — as a feature that “increases snow accumulation.” It states that “simple roofs with steep slopes shed snow most easily” while “roofs with geometric irregularities and obstructions collect snow drifts in an unbalanced pattern,” and it names “saw-tooth roofs” among those geometries. A butterfly roof is not a sawtooth roof, and nothing here claims FEMA said it was. The sentence that fits this shape is the valley one: two opposite sloped roofs meeting, and more snow collecting where they do. On a butterfly that intersection runs the whole length of the building.
Three of the failure routes FEMA lists land on this shape at once. Unbalanced load from drifting and sliding snow, because both planes slide their snow into the same trough. Rain-on-snow, because a saturated drift in a trough has nowhere to shed to. And snow melt between events: FEMA notes that if “the roof drainage system is blocked, improperly designed or maintained, ice dams may form,” adding that “on flat or low slope roof systems, snow melt may accumulate in low areas on roofs, creating a concentrated load.” The trough is the low area.
The practical consequence is that the winter drainage path has to be thought about as a path, not as an opening. An internal drain and leader running inside the heated envelope behaves very differently from a scupper through an exposed parapet, which is outside in the weather for its whole length. Which is right for a given building is a design decision made with the climate, the assembly, and the interior layout in front of you. This page does not have any of those.
The heat-loss half of the mechanism — why a roof melts snow from underneath in the first place, and why that is an air-sealing problem rather than a roofing one — is set out at ice dams. The load half is at heavy snow and ice.
Debris in the valley is not cosmeticSection link
On most roof shapes, clearing the gutters is housekeeping. On this one it is the maintenance the structural argument depends on, and it is the item most likely to be quietly dropped after the first few years.
A butterfly roof collects, in one line, everything that lands on it: leaves, needles, seed pods, roofing granules, windblown grit, tennis balls. It has to, because that is the same property that makes it collect water. There is no version of this shape that sheds debris to an edge, because there is no edge.
The consequence is that the drain, the sump, and the overflow inlet are all sitting in a debris trap, and the whole safety argument for the roof — that if the primary path blocks, the secondary path runs — depends on the secondary path not being blocked by the same debris. That is a maintenance commitment with a schedule, and the schedule is often the condition a warranty is written against.
What that looks like in practice: the trough and both drain inlets checked and cleared at least at the start and end of the wet season and after any storm that brings debris down; the discharge point observed during rain to confirm water is actually coming out; the terminations and the transition detail looked at once a year; and every one of those visits written down with a date. The documentation is not bureaucracy. In a warranty argument it is the evidence.
What the shape does to the price, without inventing a numberSection link
This page publishes no dollar figure for a butterfly roof, because no honest one exists at the level of a shape. What can be said is which mechanisms move the number, and in which direction.
A roofing price is built from area, material, tear-off, disposal, access, complexity, and market. Butterfly geometry enters through complexity, and it does so in ways that are easy to describe and impossible to generalise into a national figure.
- Two systems instead of one. The planes and the trough are usually different roofing systems with different material costs and, often, different crews. That is a scheduling cost as well as a material one.
- Drainage as designed work. Drains, sumps, leaders, conductor heads, and an overflow are engineered components with a drawing behind them. On a gable, the drainage design is a gutter sized by habit.
- Building the fall. If the trough gets its slope from a tapered layer, that is material volume nobody sees, priced by the board foot. If it gets it from the framing, it was priced years ago and cannot be changed now.
- Terminations and transitions. Every foot of the trough has two upturns and two edges. Detail work is the slowest labour on any roof and this shape is mostly detail.
- Access and staging. Working in a trough between two rising planes is slower and needs more fall protection than working on an open slope.
- Maintenance over the life of the roof. Rarely quoted, and rarely nothing. This shape has a standing obligation that a gable does not, and it runs for as long as the roof does.
If you want the money conversation itself, it lives at roof cost, and the basis behind every figure on this site is published at the cost methodology. If you are comparing proposals, the line items to force into the open are in the list above the source list, and the quote comparison guide explains how to normalise scope before price.
What changes this on a real buildingSection link
Eight things that change what a butterfly roof means on a specific building.
- Slope and drainage
Slope on a butterfly roof is two separate questions that get confused with each other. The planes have a pitch, and that pitch decides which coverings may be used on the planes and how fast water arrives at the trough. The trough has its own longitudinal fall, and that fall decides whether the water leaves. A steep butterfly with a level trough drains no better than a shallow one.
Minimum design slopes are set covering by covering in whichever code edition your jurisdiction adopted, and the figures quoted on this page belong to the City of Seattle. Confirm your own edition, amendments, and effective date with your authority having jurisdiction.- Moisture and ventilation
Everything that goes wrong with a butterfly roof goes wrong at the trough, and most of it is a flashing and termination story rather than a field-membrane story. The adopted residential code read for this page requires flashing “wherever there is a change in roof slope or direction,” which on this shape describes the entire length of the valley. Read where roofs actually leak; almost none of it is in the middle of a plane.
- Structural weight
Water is heavy and a blocked trough accumulates it in one place. Standing water an inch deep weighs about 5.2 pounds per square foot, which is arithmetic anyone can do. What that means for a particular deck, in a particular framing arrangement, at a particular span, is not arithmetic. IIBEC describes the failure mode plainly for roofs lacking sufficient framing stiffness: “increasingly larger deflections caused by the continued accumulation of rainwater are large enough to overload the structure and result in a roof collapse.”
Nothing here is a structural, drainage-design, or snow-load determination for any building. Rain and snow loads, deck stiffness, ponding instability, and drain sizing are decided by a licensed design professional working to the standard your jurisdiction has adopted.- Climate
In cold climates the trough is where snow ends up and where melt refreezes. FEMA’s snow guidance identifies a valley as an “intersection at two opposite sloped roofs” that “increases snow accumulation,” and lists “saw-tooth roofs” among the geometries that “collect snow drifts in an unbalanced pattern.” It is the valley sentence, not the sawtooth one, that describes this shape: a butterfly roof is that intersection, running the length of the building. See the heavy snow and ice guide and freeze-thaw.
- Maintenance
A butterfly roof asks for a maintenance schedule in a way that a gable does not. FEMA’s list of causes of snow-related structural failure includes drainage that is “blocked, improperly designed or maintained,” and its practical instruction is to keep snow away from drain downspouts. The equivalent instruction in every other season is that the trough and the drain are inspected and cleared, by someone equipped to be up there.
- Code and jurisdiction
There is no nationwide building code for site-built construction in the United States. Every figure quoted on this page comes from one named jurisdiction’s adopted code, with its edition and effective date recorded in the source list, because a code figure with a jurisdiction attached can be checked and one without cannot. Drain and overflow sizing sits partly in the adopted plumbing code, which is a separate document again.
Confirm the adopted edition, its amendments, its effective date, and what is actually required on a permit with your authority having jurisdiction. The Seattle provisions quoted here are the law in Seattle and nowhere else.- Access and site conditions
The trough is the working surface of a butterfly roof and it is the least forgiving place on it to work: wet more often than anywhere else, often the only place a person can stand without sliding, and directly over the room you least want a foot to go through. Design for access — a hatch, a safe route, a walkway pad — is part of the roof, not an extra.
- Fire
Nothing about butterfly geometry sets a fire classification. What the shape does contribute is a debris trap: the trough collects leaves and needles in exactly the way an ordinary roof plane does not, and accumulated organic debris in a valley is an ember target as well as a drainage problem.
Fire classification — Class A, B, or C — applies to a tested roof assembly, deck and underlayment and covering together, not to a covering in isolation and never to a roof shape. Wildland-urban interface requirements are jurisdictional.
What a warranty will and will not do about a valley that pondsSection link
This is the single most important warranty question on a butterfly roof, and it is one most buyers do not think to ask until the water is already standing.
- Ponding is where the warranty argument lands, and the trough is where ponding happens
Membrane manufacturers write their own warranties, and standing water is one of the conditions those documents address — in terms that vary from one manufacturer to the next, which is why this page will not tell you what yours says. NRCA’s position is that “ponding water for short durations is unavoidable and considered acceptable” while water lasting more than 48 hours can harm the assembly. Whether any particular document treats your trough as acceptable short-duration ponding or as an excluded condition is a question about that document. Ask for the warranty text before the system is chosen, not after.
- Drainage design is usually somebody else’s responsibility
A covering warranty is about the covering. The number of drains, where they are, how big the leaders are, and whether the trough has any fall are design decisions made before a membrane was selected. A roof that ponds because it was drawn to pond is not ordinarily a manufacturer’s problem, and it may not be the installing contractor’s either. Establish in writing who owns that decision.
- Maintenance conditions have teeth on this shape
Warranties commonly condition coverage on inspection and maintenance. On a gable, an unmet maintenance condition is a technicality. On a butterfly roof, where a blocked drain is the central failure mode, it is the argument. Keep dated records of every inspection and every clearing. Read how roofing warranties actually work.
Repairability
A butterfly roof repairs in two halves. The planes are ordinary: whatever covering is on them can be repaired the way that covering is normally repaired, and neither plane has valleys, hips, or dormers unless the design added them. The trough is not ordinary. It is a membrane detail with terminations, a drain flashing, and an overflow, and repairing it competently is a low-slope trade skill rather than a steep-slope one.
That split is worth knowing before you hire. A contractor who does excellent shingle work may have no business in the trough, and one who does excellent residential low-slope work may not want the planes. On many butterfly roofs the honest answer is two trades, sequenced, with one of them owning the transition.
A warranty is a contract between a reader and whoever wrote it. What it covers, what voids it, whether it transfers, and how it is enforced are set by that document and by the law where the reader lives. Read the actual warranty for the product and the installer in front of you — not a summary of one, including this one.
Questions to ask an installerSection link
These are questions about your building, not about products. A contractor who answers them specifically has looked at the roof; one who answers them generally has not.
Where exactly is the low point, and how much fall does the trough have along its length?
This is the question the whole roof turns on, and it has a numeric answer: so many inches of fall over so many feet. “It drains fine” is not an answer. If nobody knows, that is worth finding out before anything is specified.
Is there a secondary overflow, where does it discharge, and how high is its inlet above the low point?
An overflow that discharges into the same leader as the primary drain is not an overflow. One that discharges somewhere nobody ever looks is only half of one — its purpose is to be a visible nuisance. In the adopted code quoted on this page the inlet is set 2 inches above the low point of the roof served.
Is the primary drain in a sump, and how wide is that sump?
NRCA suggests a drain sump width of at least the drain bowl’s diameter plus 24 inches, so a 12-inch bowl gets a sump not smaller than 36 by 36 inches. A drain set flush in a flat trough collects a ring of standing water around its clamping ring for the life of the roof.
What system is going in the trough, and how far up the planes does it turn?
You are looking for a waterproofing membrane with terminations well above any level the water can stand at, not a water-shedding covering carried down into a place where it can be submerged. Ask for the termination height in inches above the low point.
What happens to this roof when the drain blocks in a storm — walk me through it.
A good answer describes a sequence: water rises, reaches the overflow, discharges visibly, somebody sees it, the drain gets cleared. A bad answer changes the subject to product quality. The point of the question is to find out whether the second path exists at all.
How do I get to the trough to inspect and clear it, safely, twice a year?
If the answer is “you climb up there,” the roof has a maintenance plan that will not survive contact with reality. Access, a walkway route, and who is coming to do it are part of specifying this shape.
In winter, what is the plan for snow and ice in the valley?
In a cold climate this is a design answer, not a product answer. A contractor who has thought about drifted snow in the trough and about a melt path that does not simply refreeze at the outlet is a contractor who has built one of these before.
Require these in writing
- The fall of the trough, in inches of drop over its stated length, and how that fall is achieved — warped planes, tapered build-up, or purpose-framed gutter.
- The number, size, and location of primary drains or scuppers, and the size and route of every leader.
- The secondary overflow: type, size, inlet height above the roof low point, and where it discharges.
- Whether each primary drain sits in a sump, and the sump’s plan dimensions and slope.
- The membrane system in the trough, by name, with the height its terminations reach above the low point.
- The covering on each plane, stated separately from the trough system, and the detail at the transition between them.
- Where flashing occurs at every change in roof slope or direction, with the metal type and thickness.
- The maintenance schedule the warranty conditions coverage on, in writing, with who is expected to perform it.
- A written statement of what was not inspected and why.
Misconceptions and failure modesSection link
Common misconceptions
Common belief
A butterfly roof is just two shed roofs pointed at each other, so it drains like two shed roofs.
What is actually true
The planes drain like shed roofs. The valley does not drain like anything, because two planes falling towards each other meet along a level line. On a shed roof the water reaches a low edge and leaves. On a butterfly it reaches a low line and stops. Everything after that is design.
Common belief
The valley slope is the same as the roof slope.
What is actually true
It never is, on any roof. NRCA states that “where two roof areas of equal slope intersect to form a valley, the resulting valley slope is less than that of the two adjacent roof surfaces,” and gives the worked case: two 4:12 planes produce a valley of “only about 3:12.” A butterfly is the limiting case of the same rule, and the limit is zero. The arithmetic is set out in the geometry section above.
Common belief
The overflow is an upgrade you can decline to save money.
What is actually true
In the two adopted codes read for this page it is a requirement wherever the roof perimeter construction extends above the roof such that water is entrapped if the primary drains allow buildup. Seattle’s building code goes further on sizing: scuppers must be sized “to prevent the depth of ponding water from exceeding that for which the roof was designed,” and “the flow through the primary system shall not be considered when locating and sizing scuppers.” The secondary path is sized as though the first one does not exist.
Common belief
Standing water in the valley is normal for this shape.
What is actually true
Brief standing water is normal on any low-slope roof; NRCA’s criterion for proper slope is “that there be no ponding water on the roof 48 hours after a rain during conditions conducive to drying.” A trough that is still holding water two days later is not expressing its character. It is telling you the fall, the drain, or the maintenance is not doing its job.
Common belief
A steeper butterfly drains better.
What is actually true
Steeper planes deliver the same volume of water to the same trough faster. The volume is set by the roof’s footprint, not by its pitch — which is why rainwater-harvesting guidance computes yield from the plan area and states that “regardless of the pitch of the roof, the effective collection surface is the area covered by collection surface.” Steepening the planes raises the peak rate at the trough without widening the exit.
Common belief
It is a bad roof shape.
What is actually true
It is an unforgiving one, which is not the same thing. The geometry does real work: tall outer walls that can carry high glazing, a low centre that reduces bulk on a constrained site, and a single collection point that is a genuine advantage if you want the water. What it does not tolerate is being detailed like an ordinary roof and then left alone.
How it actually fails
- The primary drain blocks and the trough fills
- Leaves, needles, a ball, ice, or construction debris. The trough collects them by design, because it collects everything by design. With no secondary path the water rises until it finds a termination, a lap, or a fastener.What you can see: From the ground after rain: nothing coming out of the downspout or leader that serves the trough while it is obviously raining. From inside: a stain appearing along a line rather than at a point, because a filling trough leaks along its length.
- The trough was never given any fall
- Two planes falling towards each other produce a level intersection, so unless the design warped the planes, tapered the build-up, or framed a falling gutter, there is no slope along the trough at all. Nothing about this is visible from the street, and it is not corrected by a better membrane.What you can see: Water standing along the whole length of the trough rather than pooling at one end. A tide line of silt and granules across the full width. Vegetation growing in the valley.
- The termination is too low and the pond reaches it
- The membrane is turned up the planes far enough for a normal day and not far enough for a blocked drain. Once the water level passes the top of the upturn, the waterproofing has ended and the assembly behind it is doing the work.What you can see: Usually invisible until it happens, then unmistakable: a leak that appears only during heavy or prolonged rain and stops entirely between events.
- Snow drifts into the valley and melt refreezes at the outlet
- FEMA identifies the valley as an intersection that “increases snow accumulation,” names sawtooth geometries among those that collect unbalanced drifts, and describes snow melt between events as forming ice dams where drainage is blocked or badly maintained, with melt accumulating in low areas on low-slope roofs as a concentrated load. A butterfly trough is all three conditions in one place.What you can see: Ice visible in the valley from a window or from across the street. Icicles hanging at a scupper or leader. Interior leaks that appear during a thaw rather than during a storm. See the ice dam guide for the heat-loss half of the mechanism.
- A steep-slope covering was carried into the trough
- Shingles, shakes, and tiles shed water down a slope by lap. In a trough that can stand water, laps are below the water line and the mechanism has nothing to work with. The adopted residential code read for this page allows asphalt shingles only on slopes of 2:12 or greater; the trough is often below that even when the planes are not.What you can see: Sealant or mastic smeared along the valley instead of a flashed or membrane detail. Coursing that runs down into the valley rather than stopping at a metal or membrane edge. Repeated repairs at the same few feet.
- The overflow was installed but discharges somewhere nobody looks
- An overflow that runs into a concealed leader, a side yard nobody walks past, or the same discharge point as the primary drain protects the structure and tells no one. The blockage then stays blocked, and the roof spends months in its emergency condition.What you can see: A staining trail below a scupper that nobody has ever noticed. A secondary leader that shares a hopper with the primary. Ask where the overflow goes; if the answer takes more than a sentence, it is in the wrong place.
Sources and further readingSection link
Understanding Roofing / Published
Scope and limitations
- It cannot tell you what your jurisdiction requires.
- Every code figure here belongs to the City of Seattle’s adopted 2021 codes and is quoted as a worked example of how such a requirement is written, not as a national rule.
- Confirm your own adopted edition, amendments, and effective date with your authority having jurisdiction.
- It does not size a drain, a leader, a scupper, or an overflow, and it does not establish a design rainfall intensity for any address.
- Those depend on precipitation-frequency data and on the plumbing code your jurisdiction adopted, and they are design work.
- It makes no structural, drainage-design, or snow-load determination.
- The weight-of-water and rate-of-rise arithmetic here is arithmetic about one hypothetical roof; whether any real deck can carry a pond is a question for a licensed design professional looking at that building.
- It publishes no cost figure or range.
- Butterfly-roof pricing turns on the trough system, the drainage design, access, and the number of details, and no transparent national dataset separates the shape from the covering, the market, and the building.
- It publishes no service-life range.
- On this shape, service life is decided by drainage and maintenance far more than by the membrane, and a number attached to a product would be decoration.
- It publishes no architectural history and no daylighting or energy figure.
- The clerestory benefit described here is a statement about the geometry — high outer walls can carry high glazing — and not a measured performance claim.
- It cannot tell you whether a warranty responds to ponding on your roof.
- That is governed by the specific warranty document and by the law where the building is, and warranty and contract questions on this site are flagged for legal review.
- It cannot tell you what is happening on your roof right now.
- Distinguishing a design problem from a maintenance problem on a particular building takes someone standing on it who is equipped to be there.
2021 Seattle Residential Code, Chapter 9 — Roof Assemblies (adopted law, City of Seattle)
Seattle Department of Construction and Inspections — the adopting jurisdiction / 2021 edition, in effect in Seattle since 15 November 2024
Read verbatim: R903.4, “Unless roofs are sloped to drain over roof edges, roof drains shall be installed at each low point of the roof”; [W] R903.4.1, a Washington State amendment, requiring secondary emergency overflow drains or scuppers where the roof perimeter construction extends above the roof such that water will be entrapped if the primary drains allow buildup for any reason, with overflow drains the same size as the roof drains and their inlet flow line 2 inches above the low point of the roof, or overflow scuppers three times the size of the roof drains with a minimum opening height of 4 inches in the adjacent parapet walls, the inlet 2 inches above the low point of the roof served, sizing to Sections 1101 and 1103 of the Uniform Plumbing Code, and discharge to an approved location; R903.2.1, flashing installed at wall and roof intersections, wherever there is a change in roof slope or direction and around roof openings, with metal flashing not less than 0.019 inch; R903.3, parapet walls coped with noncombustible weatherproof material; R905.2.2, asphalt shingles used only on slopes of 2:12 or greater, with double underlayment from 2:12 up to 4:12; the 1/4:12 design slope for drainage required of built-up (R905.9.1, with 1/8:12 for coal tar), modified bitumen (R905.11.1), thermoset single-ply (R905.12.1), thermoplastic single-ply (R905.13.1), sprayed polyurethane foam (R905.14.1) and liquid-applied roofing (R905.15.1); and R908.1 Exceptions 1 and 2, that reroofing is not required to meet the 1/4:12 minimum design slope, and that recovering or replacing an existing roof covering does not require the secondary drains or scuppers of R903.4.1 to be added, for roofs that provide positive roof drainage.
This is the law in the City of Seattle and nowhere else. It is the 2021 International Residential Code as amended by Washington State (sections marked [W]) and by Seattle, so it is neither pure model text nor transferable to another address. It is used here because a code figure with a named jurisdiction, edition, and effective date can be checked, and one without them cannot.
2021 Seattle Building Code, Chapter 15 — Roof Assemblies and Rooftop Structures (adopted law, City of Seattle)
Seattle Department of Construction and Inspections — the adopting jurisdiction / 2021 edition, in effect in Seattle since 15 November 2024
Read verbatim: 1502.1, that design and installation of roof drainage systems shall comply with Section 1611 of the code and Chapter 11 of the Uniform Plumbing Code; 1502.2, secondary (emergency overflow) roof drains or scuppers required where the roof perimeter construction extends above the roof such that water will be entrapped if the primary drains allow buildup for any reason; 1502.3, that where scuppers are used for secondary drainage “the quantity, size, location and inlet elevation of the scuppers shall be sized to prevent the depth of ponding water from exceeding that for which the roof was designed,” that scuppers “shall not have an opening dimension of less than 4 inches (102 mm),” and that “the flow through the primary system shall not be considered when locating and sizing scuppers”; 1503.2.1, flashing at wall and roof intersections and wherever there is a change in roof slope or direction; 1507.2.2, asphalt shingles only on slopes of 2:12 or greater; and the 1/4:12 design slope for drainage required of built-up (1507.10.1), modified bitumen (1507.11.1), single-ply membrane (1507.12.1), sprayed polyurethane foam (1507.13.1) and liquid-applied (1507.14.1) systems.
Adopted law in Seattle only, and the building code rather than the residential code — it governs buildings within the scope of the International Building Code, not detached one- and two-family dwellings, which fall under the Seattle Residential Code cited above. Its ponding-depth and structural provisions sit in Section 1611, in Chapter 16, which was not read for this page; no structural claim here rests on it.
The Seattle construction codes are being updated (notice of the 15 November 2024 effective date)
Seattle Department of Construction and Inspections / 5 September 2024
That the Seattle construction codes were updated to the 2021 editions — building, residential, existing building, energy, mechanical, plumbing, fuel gas and fire — with an effective date of 15 November 2024, and that until that date applicants could use either the 2018 or the 2021 Seattle codes.
An official announcement of an effective date, not the code text itself, and specific to Seattle.
Roof slope guidelines (Professional Roofing, August 2018)
National Roofing Contractors Association — Mark S. Graham, vice president of technical services / August 2018
That “for steep-slope roofs where two roof areas of equal slope intersect to form a valley, the resulting valley slope is less than that of the two adjacent roof surfaces,” with the worked example that “where roofs with a 4:12 slope intersect at a valley, the valley’s actual slope is only about 3:12”; that for low-slope roofs where a tapered cricket or saddle creates a valley the valley’s slope will be less than that of the cricket or saddle, and that “some ponding water along cricket and saddle valleys typically will occur and should be anticipated”; the definition of positive roof drainage as the condition in which consideration has been made for all loading deflections on the roof deck and additional slope has been provided to ensure drainage within 48 hours of precipitation; and NRCA’s recommended slopes of 4:12 or more for asphalt shingle, tile, metal shingle, slate and wood systems and 1/2:12 or more for structural metal panel systems.
Best-practice guidance from a trade association of roofing contractors, not adopted law. Its code references are to the 2018 IBC and IRC, earlier editions than the adopted code quoted elsewhere on this page. Its published 4:12 example is used here to check a geometric derivation, not as a source for the derivation itself.
Still water runs deep (Professional Roofing, July 2012)
National Roofing Contractors Association — Jason Wilen, AIA, NCARB, CDT, RRO / July 2012
That ponding water for short durations is unavoidable and considered acceptable by NRCA and membrane roof system manufacturers, while ponding in excess of 48 hours can be detrimental to the roof assembly; and NRCA’s criterion for judging proper slope for drainage, “that there be no ponding water on the roof 48 hours after a rain during conditions conducive to drying.”
Trade-association guidance, not adopted law, and an older article whose code section references belong to superseded editions. It establishes a durability criterion, not a requirement anywhere, and it says nothing about what any individual manufacturer’s warranty covers.
Drain sump details (Professional Roofing, September 2025)
National Roofing Contractors Association — Maciek Rupar, director of technical services / September 2025
That a drain sump is an intentional depression around a roof drain or scupper that promotes drainage and “reduce[s] the potential for localized ponding that could be caused because of a buildup of materials and a drain clamping ring”; that “NRCA recommends drain sumps at all internal drain locations” and that they may not be necessary where tapered insulation provides adequate slope; that NRCA suggests a drain sump width of at least the drain bowl’s diameter plus 24 inches, so that a 12-inch bowl gets a sump not smaller than 36 by 36 inches; that the sump slope should be greater than the roof slope; that where primary and overflow drains share a common sump NRCA suggests a minimum edge-to-edge spacing of 12 to 18 inches; and NRCA’s caution against round or deeply recessed, sharply sloped sumps, which can wrinkle the membrane at the transition.
Trade-association best practice, not adopted law and not a design determination for any roof. The dimensions are suggestions keyed to NRCA’s own manual, and the article is written about membrane roof systems on low-slope roofs rather than about butterfly geometry specifically.
Secondary Drainage and Ponding Requirements in the IBC and IEBC
IIBEC (International Institute of Building Enclosure Consultants) — Wanda Edwards, PE / 28 November 2017
That secondary (emergency overflow) roof drains or scuppers are required where the roof perimeter construction extends above the roof so that water would be entrapped if the primary drains fail; that “if adequate drainage is not provided, structural damage—and, in the worst cases, roof collapse—can result”; the model-code definition of positive roof drainage as drainage within 48 hours of precipitation after consideration of all loading deflections of the roof deck; that ponding instability occurs in roofs lacking sufficient framing stiffness, where “increasingly larger deflections caused by the continued accumulation of rainwater are large enough to overload the structure and result in a roof collapse”; that a susceptible bay includes a roof with slope less than 1/4 inch per foot; and that roofs must be designed to sustain the load of rainwater that accumulates if the primary drainage for that portion is blocked.
Professional-institute commentary on the 2015 International Building Code and International Existing Building Code. It is not adopted law anywhere, its section numbers are not the ones a residential reader will meet, and nothing in it is a structural determination for a specific building.
FEMA Snow Load Safety Guidance — the two-page flyer, published under the same FEMA P-957 number as the full Snow Load Safety Guide
Federal Emergency Management Agency, Building Science
The list of warning signs of overstress conditions during a snow event, including sagging roof members, popping, cracking and creaking noises, doors and windows that can no longer be opened or closed, cracks in walls, severe roof leaks, and “excessive accumulation of water at nondrainage locations on low slope roofs”; the instruction that where any of those signs are observed “the building should be promptly evacuated and a local building authority and/or a qualified design professional should be contacted to perform a detailed structural inspection”; that a valley is an “intersection at two opposite sloped roofs” that “increases snow accumulation”; that structural failure may be linked to unbalanced snow load from drifting and sliding snow, to rain-on-snow, and to snow melt between events where “the roof drainage system is blocked, improperly designed or maintained,” with melt accumulating in low areas on flat or low-slope roofs as a concentrated load; that “simple roofs with steep slopes shed snow most easily” while roofs with geometric irregularities — including “flat roofs with parapets, stepped roofs, saw-tooth roofs” — collect drifts in an unbalanced pattern; that snow removal is a dangerous activity that should only be done by qualified individuals; and the instruction to keep snow away from drain downspouts.
This is FEMA’s two-page guidance flyer, which summarises the Snow Load Safety Guide and points readers to the full Guide. FEMA carries the same P-957 number on both items, so the number alone does not identify which document a citation means: this card is the flyer, and the full Guide — Risk Management Series, January 2013 — is the separate, much longer document cited on this site’s gambrel roof page. The flyer contains no snow-load figures, drift calculations, or design provisions; the full Guide was not read for this page and nothing here is attributed to it. It is federal guidance rather than adopted law. fema.gov returns 403 to automated requests, so the copy read was the FEMA-branded PDF served by the Town of Yarmouth, Massachusetts, and that municipal copy is the URL given above precisely because it is the document that was actually opened. FEMA publishes both items at fema.gov.
The Texas Manual on Rainwater Harvesting, Third Edition (2005)
Texas Water Development Board / Third Edition, 2005, Austin, Texas
That “in theory, approximately 0.62 gallons per square foot of collection surface per inch of rainfall can be collected”; that “the collection surface is the ‘footprint’ of the roof… regardless of the pitch of the roof, the effective collection surface is the area covered by collection surface (length times width of the roof from eave to eave and front to rear)”; that if only one side of a structure is guttered, only the area drained by the gutters counts; that rainwater is lost in practice to first flush, evaporation, splash-out or overshoot from gutters in hard rains, and possibly leaks, and that rough collection surfaces are less efficient; that spillage occurs when the flow-through capacity of a filter-type roof washer is exceeded and rainwater is lost as overflow once tanks are full; and that most installers assume a collection efficiency of 75 to 90 percent, with the manual’s own worked example using 85 percent.
A state-agency planning manual for rainwater harvesting in Texas, not a roofing or drainage design document. It is cited here for the runoff arithmetic and the footprint-not-pitch point, both of which are general. Its potability, filtration, and tank-sizing guidance is specific to harvesting systems and to Texas rules, and none of it is relied on here.
Precipitation Frequency Data Server (PFDS)
National Oceanic and Atmospheric Administration — National Weather Service, Office of Water Prediction, Hydrometeorological Design Studies Center
That the PFDS is “a point-and-click interface developed to deliver NOAA Atlas 14 precipitation frequency estimates and associated information,” with estimates and confidence intervals available as tables or graphs, and supplementary materials including grids of estimates, temporal distributions of heavy rainfall, time series at observation sites and cartographic maps — that is, that a location-specific design rainfall figure exists and is obtained from published data rather than assumed.
A data service, not a code. It does not tell you which return interval or duration your jurisdiction’s adopted plumbing code requires you to design to, it does not size any drainage component, and the coverage and vintage of NOAA Atlas 14 vary by region. The 3 inches per hour used in the worked example on this page is an arbitrary round number chosen for arithmetic and is not taken from this source.
Residential Fall Protection — guidance
U.S. Occupational Safety and Health Administration
That “falls are the leading cause of death for workers engaged in residential construction,” and that workers engaged in residential construction six feet or more above lower levels “must be protected by conventional fall protection (in other words, guardrail systems, safety net systems, or personal fall arrest systems).”
An occupational-safety standard for employers and workers, not a rule that governs a homeowner on their own house. It is cited here as evidence of how seriously the hazard is treated for trained, equipped people.
Learn About Asbestos
U.S. Environmental Protection Agency
That roofing shingles are among the products in which asbestos may be found, and that “asbestos fibers may be released into the air by the disturbance of asbestos-containing material during product use, demolition work, building or home maintenance, repair, and remodeling,” exposure generally occurring only when the material is disturbed or damaged so that particles and fibres are released.
General federal guidance on asbestos, not roofing guidance. It gives no cut-off year after which roofing materials are asbestos-free and it cannot tell you whether any particular roof contains asbestos. Where this page mentions roofs predating about 1990 it is naming a reason to test before disturbing old felts, mastics and flashing cements, not a threshold that settles anything: no age proves a material is or is not asbestos-containing, and testing by someone qualified to sample and interpret is the only way to know.