Two fields with random aperture distribution and different spatial structures are taken as models to study solute transport in fractures. One network has non-vanishing long range correlations and represents a fractal pattern. The other one has a finite correlation length and an exponential covariance function. Based on these fields, two physical fracture models were produced and used to record the movement of a coloured solute by means of a CCD camera. The pictures obtained were analyzed with

more »

... age processing methods. A front tracking algorithm shows that the growth law of the frontal variance is a power law of time with the exponent depending on the Hurst coefficient of the aperture distribution in the case of the fractal pattern, while it is a linear function of time for the case of the finite correlation length.