The heterogeneity of the local surface temperatures over an urban area, due to the geometrical complexity of the canopy and to the diversity of the thermal properties of the different materials, generates strong thermal anisotropy effects at the district scale. To simulate these effects, two complementary studies are carried out and the results compared against airborne thermal infrared measurements obtained over Toulouse (France) during the CAPITOUL experiment (http://medias.cnrs.fr/capitoul/). Both studies are based on the SOLENE software. SOLENE simulates the air-solid thermo-radiative transfers coupled with the visible and thermal infrared radiative transfers. Parameterizations of thermal inertia and heat exchange through walls are also introduced, with several layers. The computations are made with fine meshes over all the facets of the urban canopy described by a 3D model. Input meteorological data is used and simulations are performed with a 15 minutes time step over periods long enough to ensure representative thermal regimes.

SOLENE is first used to simulate the local radiative surface temperature (at 1 m resolution) of a 18 000 m˛ urban fragment of the center of Toulouse. The derived TIR images are then directly compared against airborne high resolution infrared thermography measurements, corresponding to the same urban fragment. The comparison allows to validate the parameterizations of the model and to fit the main parameters (thermal and radiative properties) over the urban district.

In the second study, simulations are made for a simplified canyon street geometry (at 1 m resolution) with the scope of simulating the TIR directional anisotropy. The geometrical characteristics (aspect ratio, street and roof widths) were determined using the Toulouse database. The simulations are performed for 18 different street orientations by 10° steps to describe all the possible directions of streets. The simulated temperature profiles are then integrated for different viewing positions and the directional temperatures are determined for 6 different classes of surfaces: roofs, walls and grounds, sunlit and shaded. Their ratio in a given viewing direction are then derived from images of the urban test area (about 3 km˛) generated using the POVRAY software. The ratios are used to weigh the temperatures of each class and to compute the resulting directional brightness temperature at the district scale. The simulated anisotropy is finally compared against the anisotropy derived from the TIR airborne measurements for 4 flights performed over Toulouse on July 15th 2004 and February 25th 2005.