The characterization of the morphology of rock joint surfaces is of great interest, in particular for the detennination of realistic shear strength of rock joints. From a geometrical point of view, a rock joint surface is characterized by irregular asperities which form polygonal profiles. In this framework, a new method devoted to morphology characterization has been developed. This new spectral analysis approach appears really suited to the natural features of discontinuities in rock masses. The method is based upon a serial decomposition with triangular functions. It is named ATM method. Experimentally the measurements of the rock joint microtopography were realized firslly in the laboratory with a three-dimensional roughness apparatus and secondly in situ with a Profilographe rule. The research work deals firstly whith the characterization and the quantitative description of the rock joint morphology at different scale. Many simulations are presented in order to validate and demonstrate the interest and quality of this new method. Some of them are devoted to the comparison willl the well known Fourier spectral analysis. In the second part of this research work, the spectral information issued from the ATM method are taken into account in order to perform a transformation of the initial three-dimensional geometry to a bidimensional one. The behaviour of this 2-D morphology is supposed to be quite simitar to the real one, notably in shear analysis (Dilatancy phenomenon). In addition, this work has led to the development of a numerical code, concerned with this new spectral analysis and named MORPHEE. This program is presented and numerical simulations are provided in order to demonstrate its capabilities. Many computations are done in order to establish it superiority by comparison with the Fast Fourier Transform. The numerical results which have been compared with the experimental ones allow to conclude that this new method is weil validated.
