The thermal emission received from a planet is a key to characterize the physical and chemical properties of its atmosphere. The spectral resolution, and therefore the spectral characterization that we can achieve for extrasolar planets is unfortunately very limited in particular for terrestrial planets. In this thesis, we study the possibility to characterize an Earth-like exoplanet by the analysis of the broadband infrared emission variability. The apparent thermal emission is indeed modulated by the seasons, the rotation of the planet, the motion and variability of atmospheric patterns and clouds, the orbital phase, and even the presence of a moon. As a reference case, we have studied the thermal emission of the Earth seen as a distant point-source, as well as the variability and the dependency of the signal on the observation geometry. We have modeled the emission of the Earth using data derived from observations and data produced by a General Circulation Models (GCMs), comparing both types of data in order to validate our simulations.As a second part of our work, we have used the GCM to model Earth-like planets that differ from the Earth by a few parameters such as different rotation rates (including tidally-locked planets), a planetary surface completely covered by ice or water, different obliquities and eccentricities. For all these virtual planets, we have studied the climate derived from the physical properties and the photometric infrared signal associated with them.The last part of the thesis is a preliminary work that no longer consider the bolometric emission but the signal from narrow spectral bands, thanks to a new generation of GCMs. Because each band probes a specific level in the atmosphere, studying how the low resolution spectrum of the planets varies by comparing the photometric variabilities between bands, we can study the dynamics, composition, distribution and evolution of the atmosphere of the planet, which it opens a yet unexplored field for the characterization of exoplanets.
