Ground mount or roof mount?

Two different code problems, not two versions of one

A rooftop array is an addition to a structure somebody else designed for a different purpose. Everything difficult about it follows from that: the roof must satisfy its own loading requirements independently and then carry the array, drift created by the panels must be accounted for, penetrations must be flashed, and attachment must find framing.

A ground-mounted array has none of those problems, because it is not on a building. It is designed under the general structural provisions as its own structure, with its own foundations, resisting wind and seismic loads on its own merits. Roof live load does not apply to it at all, because there is no roof.

That is a genuine simplification and it is the strongest argument for ground mounting. The entire body of roof-related risk that we cover elsewhere, flashing, leaks, warranty exclusions, re-roofing sequencing, simply does not arise.

What replaces it is a different set of constraints, and they are not structural. They are about fire and about land use, and both catch people out because neither is where they are looking.

The constraint people do not expect: a clear perimeter

The fire code addresses ground-mounted arrays directly, and it does two things at once.

It removes a requirement. Verbatim: "setback requirements shall not apply to ground-mounted, free-standing photovoltaic arrays". The access pathways and setbacks that a rooftop array must leave for firefighters do not apply here, because firefighters are not going to be working on a roof under it.

And it substitutes another. The same provision requires a minimum ten-foot perimeter clear of brush around the array, and a maintained surface beneath it that is either vegetative and managed or noncombustible.

That is a land constraint, and it is larger than people picture. A ten-foot clear perimeter on all sides means the footprint of the installation is substantially bigger than the footprint of the panels. On a modest lot it can be the reason a ground mount does not fit at all, and it is not something a panel layout drawing shows.

It is also an ongoing obligation rather than a one-time condition. The surface beneath must be maintained, which means someone keeps vegetation under control for the life of the system, in a location where mowing is now awkward because there is an array over it. That is a real maintenance commitment and it is worth deciding who is doing it before the design is fixed.

Does a ground mount produce more? We cannot tell you

A ground-mount production advantage is asserted constantly and we could not verify it, so here is exactly what the evidence supports and where it stops.

The modelling tool most estimates come from assigns different assumed operating temperatures to the two. A roof-mounted array at typical standoffs is modelled at a nominal cell temperature of roughly 49 degrees, and an open rack at roughly 45. The documentation gives the reason: "based on the reduced air flow and thus higher operating temperature of a roof mount system".

Since photovoltaic output falls as cells warm, that implies an advantage for open mounting. The mechanism is sound and the temperature coefficient on any data sheet tells you the direction.

But that is an assumption inside a model, not a measurement. It is how the tool chooses to represent two mounting configurations, and we found no matched-system field study measuring the difference on real installations.

So the honest statement is: there is a plausible small advantage, grounded in a real mechanism, that nobody appears to have measured directly. If someone quotes you a percentage, ask where it came from.

The larger output difference is usually orientation, not temperature. A ground mount lets you choose azimuth and tilt; a roof gives you what it has. On a house with a poor roof aspect that freedom is worth far more than any thermal effect, and unlike the thermal effect it is something you can calculate for your own site.

When each one is actually right

Roof mount is the default for good reasons. The structure already exists, the land is already committed, there is no trenching, and it is almost always cheaper. If your roof is sound, reasonably oriented, unshaded and has years of life left, the case for anything else is weak.

Ground mount earns its cost in specific situations. When the roof does not clear the shading threshold and the ground does. When the roof is old and you would rather not sequence a re-roof around an array. When the roof faces badly and orientation freedom is worth real output. When the roof is structurally marginal. And when you have land and want a system larger than the roof supports.

The costs are real and mostly not the panels. Foundations, a structure that resists wind on its own, and a trench and conduit run back to the house. That last one scales with distance and is the item most often discovered after a design is agreed.

And there is a maintenance trade in both directions. A ground mount is reachable, which makes cleaning, inspection and repair ordinary tasks rather than roof work. It is also accessible to people, animals and mowing equipment, and it obliges you to keep the ground beneath it managed for the life of the system.

We are not giving you a cost comparison, because it depends on span, ground conditions, trench length and local loading requirements, and no primary source we could reach publishes one.

The practical items that only exist on the ground

Several parts of a ground-mount project have no rooftop equivalent, which is why quotes are hard to compare line by line.

The trench. Conductors have to reach the house, usually underground, in conduit, at a depth your jurisdiction specifies. The cost scales with distance and with what the trench has to cross: a lawn is straightforward, a driveway or a mature root system is not. This is the item most often discovered after a design is agreed.

Ground conditions. Foundations depend on what is under the site. Rock, a high water table, fill, or a frost depth requirement all change the foundation design, and none of them is visible from a satellite image. A geotechnical assessment may be required and is worth knowing about early.

Access during construction. Getting equipment to the array location matters in a way it does not for a roof, where everything arrives through the front. A site behind a house with no side access is a different job from one beside a driveway.

Security and interference. An array at ground level is reachable, which is convenient for maintenance and also means it can be leaned on, mown into, played around and occasionally interfered with. Fencing is sometimes wanted and is another cost and another zoning question.

And the land is committed. The array plus its required perimeter occupies a part of your garden permanently. That is not a cost in the quote and it is a real one, and it is worth walking the site and marking out the actual footprint including the perimeter before deciding.