Roof Repair & Leak Detection

Water enters at a break, lands on the underlayment or the top of a rafter, and runs downhill until something interrupts it. A sheathing seam, a truss plate, a run of wiring or a light box will each stop it and let it fall, and any of them can sit a long way from where the water came in. The spot on the ceiling tells you where the path ended and nothing about where it began.

Finding the entry point is mostly an attic job. With a light and some time, the underside of the deck gives up what the ceiling hides: darkened sheathing in one rafter bay, rust on a nail shank inside that same bay, a clean water line running along the side of a rafter, insulation matted in a spot that does not line up with the stain below it. Rusted nail shanks across the whole deck are a different finding, because that is an attic that cannot breathe. Read bay by bay, the attic narrows the search on top of the roof from a whole plane to a few feet, and the few feet it points at are almost always a transition rather than open field.

Some leaks resist all of that. Two small breaches feeding one stain look like a single leak, so closing the first one does not stop it. A roof that only leaks when rain arrives from one direction looks sound on a calm day. And what presents as a leak is sometimes condensation off an attic that cannot ventilate, which wets the deck from underneath and drips exactly like rain. Confirming a fix matters as much as finding it, which is why a repaired area is worth running with a hose, low to high, with somebody watching from inside.

That water gets in at transitions rather than in the middle of a plane is the general case, and it is argued on the residential roofing page. The question on a repair is narrower. Given that the attic has pointed at one wall run, one valley or one boot, what is actually wrong with that detail, because each of them fails in its own way and the failure mode decides the fix. A valley with metal in it and a valley cut closed are not even the same repair: one is a piece of flashing, the other is a lap and a cut line that were wrong from the day they went on.

Step flashing at a wall is the clearest example. The usual story is not corrosion. It is that the flashing was left in place during an earlier re-roof instead of being replaced, so a twenty year old flashing now sits under a five year old shingle, and the failure dates from the last roofing job rather than from the weather. The attic reads that: staining that arrives at the wall line and stops there points at the wall, not at the field above it. Kick-out flashing at the bottom of the same run is the piece most often missing altogether, and when it is missing the water goes behind the siding instead of onto a ceiling, which is why that one tends to be found late and found expensive.

Boots are quicker to confirm. The rubber collar around a vent pipe splits at the top where the sun is hardest on it, and that split is usually visible from a few feet away. Chimneys and skylights take longer, because both get sealed when they should be flashed, and the correct detail is different for each. On a chimney it is counterflashing cut into a mortar joint and lapped over the base flashing, so nothing depends on a bead of caulk holding. A skylight has no masonry to cut into. It wants a sill pan below, step flashing up the sides, head flashing with a diverter above, and the curb tied into the underlayment rather than sealed to the shingles, and what usually fails on one is the head flashing or the factory seal at the glass.

New shingles won't match a roof that has been weathering for a decade, and shingle lots shift over time regardless, so a repair is usually visible from the right angle. That is cosmetic, and it is worth knowing before the work rather than after it.

A repair is the right answer more often than the industry admits. An isolated failure at a boot, a valley or a wall flashing, on a roof whose deck is dry and whose covering still has years in it, is a repair. It doesn't turn into a replacement because somebody with a ladder found it.

What changes the answer is the deck and the spread, and those are two separate readings. Sheathing that is soft across several rafter bays is the structural one, because wood that has lost its integrity will not hold a fastener and no amount of covering put back over it changes that. Staining across several bays says something narrower: water has been moving widely, which is a scope question, and whether it has been doing so for years or for one bad storm depends on whether the wood under the stain is still sound. A lifted ridge can wet several bays in an afternoon and leave a permanent mark on a deck that is otherwise fine. Shingles that lost their seal across every plane in a wind event will lie flat again and look intact while the roof is no longer wind resistant, so replacing the three that blew off treats the visible part of a roof-wide problem. And the third repair in two years on the same roof is the roof making an argument.

Scale eventually runs into the code. Once a large enough share of a roof is opened up, the work stops being a repair as far as the building official is concerned and gets permitted as replacement, brought up to current fastener and secondary water barrier requirements. That threshold is stricter in Florida than in Georgia, and it sits in the code rather than with the contractor. Worth settling before a scope is written. The replacement side of this decision is covered on the residential roofing page.