Aims. Using data from the Rosetta mission to comet 67P/Churyumov–Gerasimenko, we evaluate the physical properties of the surface and subsurface of the nucleus and derive estimates for the thermal inertia (TI) and roughness in several regions on the largest lobe of the nucleus. Methods. We have developed a thermal model to compute the temperature on the surface and in the uppermost subsurface layers of the nucleus. The model takes heat conduction, self-heating, and shadowing effects into account. To reproduce the brightness temperatures measured by the MIRO instrument, the thermal model is coupled to a radiative transfer model to derive the TI. To reproduce the spatially resolved infrared measurements of the VIRTIS instrument, the thermal model is coupled to a radiance model to derive the TI and surface roughness. These methods are applied to Rosetta data from September 2014. Results. The resulting TI values from both instruments are broadly consistent with each other. From the millimetre channel on MIRO, we determine the TI in the subsurface to be <80 JK−1 m−2 s−0.5 for the Seth, Ash, and Aten regions. The submillimetre channel implies similar results but also suggests that higher values could be possible. A low TI is consistent with other MIRO measurements and in situ data from the MUPUS instrument at the final landing site of Philae. The VIRTIS results give a best-fitting value of 80 JK−1 m−2 s−0.5 and values in the range 40–160 JK−1 m−2 s−0.5 in the same areas. These observations also allow the subpixel scale surface roughness to be estimated and compared to images from the OSIRIS camera. The VIRTIS data imply that there is significant roughness on the infrared scale below the resolution of the available shape model and that, counter-intuitively, visually smooth terrain (centimetre scale) can be rough at small (micrometre–millimetre) scales, and visually rough terrain can be smooth at small scales.

We simulate the stresses induced by temperature changes in a putative hard layer near the surface of comet 67P/Churyumov-Gerasimenko with a thermo-viscoelastic model. Such a layer could be formed by the recondensation or sintering of water ice (and dust grains), as suggested by laboratory experiments and computer simulations, and would explain the high compressive strength encountered by experiments on board the Philae lander. Changes in temperature from seasonal insolation variation penetrate into the comet’s surface to depths controlled by the thermal inertia, causing the material to expand and contract. Modelling this with a Maxwellian viscoelastic response on a spherical nucleus, we show that a hard, icy layer with similar properties to Martian permafrost will experience high stresses: up to tens of MPa, which exceed its material strength (a few MPa), down to depths of centimetres to a metre. The stress distribution with latitude is confirmed qualitatively when taking into account the comet’s complex shape but neglecting thermal inertia. Stress is found to be comparable to the material strength everywhere for sufficient thermal inertia (≳50 J m−2 K−1 s−1∕2) and ice content (≳45% at the equator). In this case, stresses penetrate to a typical depth of ~0.25 m, consistent with the detection of metre-scale thermal contraction crack polygons all over the comet. Thermal fracturing may be an important erosion process on cometary surfaces which breaks down material and weakens cliffs.