As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Almost every surface encountered in building and electrical inspection has an emissivity between 0.90 and 0.97, which is why a camera left at a default near 0.95 produces usable readings on plaster, brick, concrete, timber, matt paint, rubber, cable insulation, water and skin. The exceptions are narrow and severe: bare and polished metals run from about 0.02 to 0.10 when bright and 0.4 to 0.9 once oxidised, and anything glossy sits unpredictably below its matt equivalent. So the practical skill is not memorising a table, it is recognising the handful of surfaces that leave the cluster, because on those the default is not slightly wrong but wrong by tens of degrees.
Why most building surfaces cluster in one narrow band
The reason so many different materials land in the same place is that long wave emission depends on how the surface interacts with radiation around ten micrometres, and most non metals are poor reflectors and strong absorbers there. Organic binders, minerals, ceramics, water and biological tissue all have strong vibrational absorption in that band, and by the equality between absorption and emission at a given wavelength, strong absorbers are strong emitters.
Metals are the exception because their conduction electrons reflect long wave radiation very efficiently. That is the same property that makes them good electrical conductors, so the correlation is not a coincidence. A bright metal surface is close to a mirror in the infrared for exactly the reason it is close to a mirror in visible light.
This has a useful consequence for anyone surveying building fabric. Within one frame of plaster, timber, paint and brick, the emissivity variation is small enough that comparing one area against another is meaningful without touching the setting. The moment a pipe, a duct, a flue or a bare conductor enters the frame, that stops being true, and the two parts of the image are no longer on the same scale. The underlying quantity is explained in emissivity in thermal imaging, and the mechanics of entering a value are in setting emissivity on a thermal camera.
A table of published values and what they do to a reading
The figures below are the conventional published ranges used across the field, not measurements of any product. They are properties of surfaces, so they are knowledge rather than a specification of anything sold. The last column gives the direction and rough size of the error from leaving a camera at 0.95 while looking at something warmer than the room.
| Surface | Conventional emissivity range | Effect of leaving the camera at 0.95 |
|---|---|---|
| Matt black electrical tape | About 0.95 | None worth correcting; this is the reference target |
| Plaster and gypsum board | 0.90 to 0.95 | Reads very slightly low; negligible for survey work |
| Common brick and concrete | 0.90 to 0.95 | Reads very slightly low |
| Planed timber | 0.90 to 0.95 | Reads very slightly low |
| Matt paint, any colour | 0.90 to 0.96 | Negligible; pigment has almost no effect in this band |
| Rubber and PVC cable insulation | 0.90 to 0.95 | Negligible |
| Water and ice | 0.95 to 0.98 | Negligible, but a water surface reflects the sky strongly outdoors |
| Human skin | About 0.98 | Reads marginally low |
| Soda lime glass, viewed square on | 0.85 to 0.95 | Reads low, and the surface is opaque so you measure the glass |
| Oxidised steel | 0.75 to 0.95 | Reads low by a few degrees to tens of degrees depending on the oxide |
| Stainless steel, rolled | 0.15 to 0.35 | Reads badly low; needs a target or a contact probe |
| Oxidised copper | 0.4 to 0.8 | Reads substantially low and varies across the same bar |
| Polished copper or aluminium | 0.02 to 0.10 | Reading is dominated by reflection and should not be used |
| Anodised aluminium | 0.6 to 0.9 | Reads low; much better than bright aluminium but still needs attention |
The structure of that table is more informative than any individual row. Everything from the top down to oxidised steel can be handled by one default setting. Everything below it needs a different technique rather than a different number, because the published range itself is too wide to pick a value from with confidence.
The surfaces that break the default
Three categories account for nearly every bad reading in the field.
Bright metal. Polished copper, bright aluminium, new galvanised steel, chrome and stainless all emit a small fraction and reflect the rest. The published range for each spans a factor of three to five depending on finish, so even a careful choice of value carries large uncertainty, and the result is extremely sensitive to that choice. These surfaces need a high emissivity target applied to them or a contact measurement.
Gloss finishes on anything. A gloss paint, a varnished timber, a laminated panel or a glazed tile sits below its matt equivalent and reflects specularly, which means it can show a sharp image of a lamp or a window rather than a diffuse average of the room. The giveaway is that the apparent hot or cold patch moves when you move.
Surfaces whose finish varies within the frame. A copper bar that is bright where it was cleaned and dark where it oxidised has two different emissivities in one image, so a temperature gradient appears along a bar that is actually isothermal. This is the most misleading of the three, because it looks exactly like a real thermal finding.
Where you suspect the first category, the fix is mechanical rather than electronic. Where you suspect a joint is degraded, the dead circuit measurement is the direct evidence, and the method is covered in measuring resistance safely; a resistance figure across a connection does not care what the surface looks like.
Glass, water and painted metal behave differently than they look
Three materials are worth separating out because intuition from visible light misleads.
Glass is opaque, not transparent. In the long wave band these cameras work in, ordinary glass transmits essentially nothing, so a camera pointed at a window measures the window. You cannot survey a room from outside through the glazing, and you cannot check a sealed enclosure through a glass viewing panel unless that panel is a purpose made infrared window in a material such as calcium fluoride. The visible transparency is irrelevant.
Water is a strong emitter and a strong reflector of the sky. A water surface sits near 0.96 in the infrared, so the setting is easy. Outdoors, though, it mirrors the sky, and a clear sky has an extremely low effective radiating temperature. A pond or a flat roof puddle under clear sky can read several degrees below the air around it for that reason alone.
Painted metal is paint, not metal. Emissivity is a surface property, so a painted steel enclosure reads like paint, around 0.9, and behaves well. This is quietly useful: the painted part of an assembly is often the best available target when the conductors themselves are bare, and it is usually in reasonable thermal contact with what is inside.
Solar modules are the awkward case
A photovoltaic module combines every difficulty on this page into one target. Its front surface is glass, so the emissivity is reasonably high but the surface is also strongly reflective at grazing angles. It is installed facing an open sky, which is the coldest thing in the radiative environment, so the reflected term pulls the apparent temperature down. It is usually at an angle, which means the operator either views it obliquely, worsening the reflection, or appears in the reflection. And the quantity people actually want, cell temperature behind the glass, is not the surface being measured.
This is why instrumentation for photovoltaic work tends to measure module temperature with a dedicated sensor rather than by infrared. The Fluke 393FC clamp meter with irradiance meter states irradiance, ambient temperature and PV module temperature measurement alongside array orientation and tilt angles, with a CAT III 1500 V rating, a thin jaw for crowded combiner boxes and MC4 test leads. Measuring the module with a contact sensor removes the emissivity question and the sky reflection question in one step.
The reason temperature matters at all on a module is electrical, and that side can be measured directly. The LODESTAR LD673PV solar panel tester states maximum power point measurement up to 1000 W on a single panel, with readings for Pmax, Vmp, Voc and Isc, a Voc range of 12 to 80 V, Isc up to 35 A, reverse polarity protection and a note that it is for single panel measurement rather than series or parallel strings. Module output falls as cell temperature rises, so a direct electrical measurement captures the consequence that the infrared image can only suggest. The DC side measurement technique is covered in measuring DC voltage with a multimeter.
Infrared still earns its place on an array, but as a locator rather than a thermometer. A cell running hot relative to its neighbours stands out clearly even when every absolute number is pulled down by the sky, because the reflection applies equally across the module and cancels in the comparison.
Using the table without treating it as gospel
A few habits turn the figures above into reliable practice.
- Sort surfaces into two bins, not fourteen. High emitters above about 0.85 take the default and need no thought. Everything else needs a target, a contact measurement or a relative comparison. The precise value inside the first bin almost never changes a decision.
- Treat metal figures as warnings, not values. A published range of 0.4 to 0.8 for oxidised copper is telling you the number is unknowable from a table. Pick a technique instead of a value.
- Check for specular behaviour by moving. One step sideways separates a reflection from a real feature, and it costs nothing.
- Note what you assumed. A survey record that does not state the emissivity setting and the ambient temperature cannot be compared with a later one.
- Remember the setting applies to the whole frame. One value is in force across every pixel, so a frame containing both plaster and bare pipework is correct for one of them at most. Where it matters, take two images with two settings rather than one compromise.
The honest summary is that emissivity tables are extremely useful for the surfaces you did not need to worry about and nearly useless for the surfaces you did. Knowing which side of that line you are on is the whole skill.
Frequently Asked Questions
Does the colour of a painted wall change its emissivity?
Barely. Colour describes how a surface behaves in visible light, and a thermal camera works in the long wave infrared where most paint binders behave almost identically regardless of pigment. A white wall and a black wall in the same room read within a fraction of a degree of each other. What does matter is the finish: a high gloss surface has a lower and less predictable emissivity than a matt one of the same colour.
Why is 0.95 the usual default?
Because it sits in the middle of the band that covers almost everything a general purpose camera gets pointed at. Plaster, brick, concrete, timber, matt paint, rubber, cable insulation, water and skin all fall between roughly 0.90 and 0.98, so one value near the top of that cluster is wrong by only a couple of percent across the whole group. The default was chosen for the common case, and it fails only outside it.
Can I measure through a window with a thermal camera?
No. Ordinary glass is effectively opaque in the long wave infrared band these cameras use, so you measure the temperature of the glass surface itself. You also get a strong reflection of whatever is behind you. Pointing a camera through a window at a room, or at a building through a glazed screen, produces a picture of the glass and your own reflected surroundings.
What emissivity should I use for a solar module?
The front glass is usually taken as somewhere between 0.85 and 0.90, but the emissivity setting is the smaller half of the problem. A module facing an open sky is reflecting a very cold effective sky temperature, so the reflected term pulls the apparent reading down well below air temperature even when the setting is right. Readings on modules should be taken as comparisons between cells and strings rather than as absolute cell temperatures.
Are published emissivity tables accurate enough to rely on?
For high emissivity surfaces, yes, because the spread within the group is small and the sensitivity of the result to the value is low. For metals, no. Published figures for metals are ranges spanning a factor of ten or more depending on finish, oxidation and temperature, and the final answer is extremely sensitive to which end of the range applies. On metal, use a tape or paint target or a contact probe rather than a table.