Roof Moisture Survey: Emissivity Correction vs. Solar Loading
A thermographer surveys a low-slope EPDM roof (emissivity about 0.90) at 0600 after a clear, calm night with a 14 K drop below the previous day's high. Ambient is 8 °C, sky is clear, wind under 1 m/s. The imager is set to e=0.95 (its default). The survey finds a 3 m × 5 m region reading 4 K cooler than surrounding membrane; the client wants it marked as wet insulation. A second crew resurveyed the same area at 1400 on a sunny day with the roof surface at 52 °C and found the same region 2 K warmer than surroundings. The imager has a stated accuracy of ±2 °C or ±2% of reading and an NETD of 50 mK.
1. Is the 0600 anomaly consistent with wet insulation, and what does the e=0.95 setting do to the reported temperature?
Read the worked answer
Wet insulation has higher thermal mass and conductivity, so at night it cools more slowly than dry insulation and appears warmer in a night survey - not cooler. A 4 K cooler region at 0600 is the opposite of the classic signature and suggests a different cause (e.g., a reflective patch, ponding with evaporative cooling, or an actual membrane defect).
Emissivity: with true e≈0.90 but the imager set to 0.95, the reported temperature is biased. Radiance is interpreted as if the surface were more emissive than it is, so the instrument over-attributes the measured radiance to emission and under-attributes reflected background. Because the reflected background (clear cold sky, ~ -20 °C equivalent) is colder than the surface, the reported temperature is biased low. The magnitude is on the order of a few tenths of a kelvin to ~1 K - smaller than the 4 K anomaly but in the same direction, so it slightly exaggerates the apparent cold spot. The 4 K anomaly is well above NETD and instrument accuracy, so it is a real radiometric difference; the question is what it means physically.
2. How should the two surveys be reconciled, and what would you require before calling the area wet insulation?
Read the worked answer
The two results are not contradictory - they are the expected time-of-day reversal. Wet, high-thermal-mass insulation lags ambient: it is warmer than dry areas at night (slow cooling) and cooler than dry areas in the afternoon (slow heating). But the observed pattern is the reverse: cold at night, warm at midday. That pattern is consistent with a region of lower thermal mass / higher insulation (heats and cools faster) or with a surface-condition effect, not with moisture.
Before calling it wet insulation, require:
- Corrected emissivity (set e≈0.90, or apply a correction) and confirmation that the region is not a reflective or low-e patch (check with a contact thermometer or e-tape reference).
- A repeat night survey under the same conditions with the region appearing warmer, confirming the thermal-mass signature.
- Verification by a secondary method - nuclear moisture gauge, impedance meter, or core sample - since IR alone is not a moisture-quantification method.
- Documentation that the anomaly exceeds instrument accuracy (±2 °C) and NETD with margin, which it does.