Wind erosion process can lead to several environmental consequences: desertification, land degradation, air pollution, etc. This last one is related to particulate matter emissions from granular materials commonly found on industrial sites, such as ore and coal. The particle size distribution of these granular materials usually consist of a mixture of a wide range of diameters, which include larger particles that are non-erodible even with strong gusts of wind. The non-erodible particles play a protective role for erodible particles, paving the surface and reducing emissions. The main objective of this thesis is to estimate more accurately emissions due to wind erosion considering the influence of the pavement caused by non-erodible particles. An analytical model was proposed to quantify emissions from particle beds and stockpiles with a wide size distribution. The effects of pavement process are incorporated in the model through the decrease of the mean friction velocity on the erodible surface as the non-erodible particles accumulate. Previous works have defined a mathematical relation between the evolution of the friction velocity over the erodible surface and the geometry of the roughness elements. Nonetheless, the formulation was only valid to limited cover rates of non-erodible particles. Numerical simulations were carried out in this work to extend the formulation in order to include other cases encountered in real situations (with larger amounts of non-erodible particles). The proposed emission model describes the relationship between the minimum value of friction velocity (at which emissions cease), taking advantage of the numerical findings, and the final eroded depth of the bed, which in turn, provides the emitted mass. Wind tunnel experiments were carried out in order to better understand the pavement phenomenon and estimate emissions from a bed of particles containing a bimodal size distribution. The experimental results were also used to validate the modeling, including the global emitted mass and the final characteristics of the bed surface. A good agreement was found between experimental and modeled results for the global emissions and the bed eroded depth. The erosion model was extended for application in stockpiles. In this case, the erodibility of the particles is more complex as the friction velocity and the threshold conditions are not spatially homogeneous. The idea of the model was to subdivide the pile in isosurfaces in which the threshold conditions and the friction velocity are constant and then treat each one of these areas as a different source where the emission model can be applied. Wind tunnel experiments were carried out in order to estimate emissions from a sand pile containing a bimodal size distribution. The modeled and the experimental results were compared for the configuration of an isolated stockpile and a good agreement was found between the estimated and the measured emitted mass. The impact of the presence of a building and a successive parallel stockpiles on the overall particles emission was also evaluated. Wind tunnel experiments and numerical simulations were carried out for several configurations evaluating the effects of: (i) main wind flow orientation, (ii) wind flow velocity, (iii) gap between the obstacle and (iv) amount of non-erodible particles. It was found that the flow interferences between the obstacles increase emissions. Therefore, all wind perturbations have a significant impact and have to be accounted in dust emission estimation and modeling.