When LiDAR meets thermal: spotting building defects no human can see

A LiDAR scan tells you what a building is. A thermal scan tells you what the building is doing. On their own, each is useful. Used together, they catch a class of defect that no walkthrough — however meticulous — will ever surface in time.

What LiDAR sees

LiDAR captures geometry to a few millimetres. On a finished façade, it tells you the panel alignment, the gap consistency, the verticality, the deviation from the design surface. On structure, it tells you slab flatness, column plumb, and any drift accumulated through the pour sequence.

What LiDAR does not see: anything behind the surface. A wall is a wall. The cavity behind the cladding, the insulation pocket inside the panel, the pipe run behind the drywall — they are invisible to a geometric scan.

What thermal sees

A thermal scan is the inverse. It sees nothing about geometry — but it reveals temperature differentials across a surface, which in a building context tracks energy loss, moisture content, and density discontinuities.

A missing batch of insulation shows up as a cold patch in winter or a hot patch in summer. A water ingress trail behind cladding shows up as a linear cold streak that does not match the panel grid. A void in a filled-in column shows up as a temperature anomaly the structural drawing cannot explain.

Why the combination is the unlock

Each instrument alone is interpretable, but ambiguous. A cold patch on a thermal scan could be insulation. It could be a moisture trail. It could be a shadow from the morning sun an hour earlier.

Thermal imaging on its own gives you a list of "things that look funny". Pairing it with the LiDAR geometric record turns those into a list of defects with locations and physical explanations.

When you align the two — same coordinate system, same time window — every thermal anomaly becomes a coordinate in the model. You can ask: "is this patch on a panel join, a window head, or in the middle of a clean panel?" The answer disambiguates the diagnosis.

A real example

On a recent commercial tower, a thermal sweep flagged a 2 m² cold patch on the south façade, two floors below a roof terrace. On its own, that pattern could have been a dozen things — the operations team would have spent days investigating.

Cross-referenced against the LiDAR model, the patch sat directly under a roof drain that had been re-routed during construction. The drain run passed inside the cavity at exactly the location of the cold patch. A visual inspection through an access panel confirmed: water was tracking down the inside of the cladding, soaking the insulation behind it.

Caught at this stage, the fix was a half-day repair. Caught a year later through a tenant complaint, it would have been a façade-wide remediation.

Where this fits in the Observance workflow

Observance scans are scheduled by floor, not by building, and the thermal sweep is layered onto the LiDAR cadence. The combined output feeds the same as-built model the project team uses for everything else — defect locations are not a separate report, they are pins on the same model surface that holds the geometry.

This is the underlying point: a defect is most expensive when it is isolated from the rest of the project's information. The LiDAR + thermal pairing makes sure no defect ever has to live in its own spreadsheet.