Abstract Solute dispersion processes in porous media remain as a subject of research due to both the limitations of available mathematical approaches, and to the need of properly reproducing the real patterns of the medium heterogeneity. Yet, building models able to accurately predict transport processes in real formations require dealing with dispersion processes. Assuming that the classical Advection-Dispersion Equation (ADE) is valid at some fine scale referred to as Representative Elementary Volume, two main and important research lines are identified. The first focused on the average ad effective dispersivity parameter that allows reproducing a solute plume development using the ADE. As a plume evolves, it flows through different heterogeneities, each of them with its own effect on the plume. These make the dispersion parameter value to increase until reaching an asymptotic value. But, different studies show that dispersivity does not reach an asymptotic value possibly due to the combined effect of different scales of hydraulic conductivity (K) heterogeneity and to the non-Fickian behavior of transport. A second line assumes the importance of properly modeling K at different scales, and that non-Fickian features can be approached by modeling heterogeneity occurring at very fine scale. Then, some mathematical approaches, such as the multirate transport formulation, allow incorporating the effects of this fine scale heterogeneity in a coarse scale, using effective values of dispersivity. Besides, the spatial variability of this parameter is not addressed in this approaches; only K variability is accounted for. Both points of view focus on a unique parameter in order to explain the deviation of real plume behavior from ADE predictions. This parameter is the heterogeneity of hydraulic conductivity, K. No attention is given to the heterogeneity of dispersivity distribution. The difficulty to estimate this parameter, and its scale dependence, has limited the number and scope of researches. Though is usual to estimate an average value for dispersitivy, these methods are not able to predict the details of local heterogeneities on the plume shape or velocity. This research is based on a solute transport intermediate scale experiment in which local and instantaneous values of dispersivity are estimated, using a laboratory tank. The available means for data monitoring, acquisition and process avoid gathering exhaustive information of the evolution of a tracer plume evolution experiment. Current researches have focused on estimating the dispersivity value using the spatial moments method or breakthrough curve analysis, mostly to verify different stochastic theories. In this research, we have built an artificial porous media, within a quasi-2D tank, with patterns of heterogeneity based on K data from a real non-Gaussian formation (MADE-2 site, Columbus Air Force base, Mississippi, USA). By controlling the head gradient between both ends of the tank (1,2 meters long, 0,40 high and 0,05 m deep) steady state flow conditions were created, and several conservative tracer tests have been conducted. The lab tank, built in plexiglas, is monitored by means of a grid of pressure transducers and by taking digital images of the evolution of dye tracer tests. By processing these digital images exhaustive information of solute concentration evolution was obtained. Results for different tracer tests have been carefully analyzed to compute both average and local dispersivities. It is found that even at this small lab scale transport anomalous behavior is observed and that average dispersivities, computed from the plume second order moments, are more dependent on the heterogeneity patterns than on the physical properties of the individual materials that make up the tank porous medium. Moreover, the successive plume images obtained have been processed over a discretization grid where, assuming the validity of the ADE, local effective dispersivities have been computed for different time steps. It is found how local dispersivity varies, in every grid cell, following a typical pattern that is significantly depends on the concentration gradient evolution and the material type within it. Although these results are consistent with some of the theoretical formulations to explain anomalous transport found in literature, provide the first available evidence, based on a controlled experiment, on how dispersivity varies in space and time, and how might be correlated with K.