Fluid flow through single rough surface is critical for many geophysical processes and engineering applications. Thus, relations between flow and surrounding fracture geometry will be investigated. Four pairs of engineered Carrara marble surfaces with different roughness properties and imposed wavelength are created. They are tested with steady-state permeability at effective stresses representative of the upper continental crust in order to study fracture transmissivity. Experimental results are compared with numerical models of flow through rough artificial surfaces in contact at an effective stress range comprised between 28 and 40 MPa. Two numerical models are used, the first one calculates the contact of both surfaces under applied effective stress and the second calculates the flow over the geometrical aperture made by both contacting surfaces. Experimental results show an exponential decrease of the transmissivity with the effective stress increase. Numerical results fit quite well experimental transmissivity results, thus complex geometries like rough fractures are nicely represented by the model. Moreover, the numerical model allows to test surfaces with only one roughness variable as Hurst exponent, roll-of wavevector, root mean square height (RMS) and imposed macro wavelength. Such roughness separated parameter analysis is quite new in the geoscience community. RMS height and the imposed wavelengths control most of the flow and may sometimes be in competition with each other in terms of transmissivity results. The tortuosity of flow which behaves like channel flow has a stronger control on the transmissivity with respect to RMS and wavelength. Such a result can be really important for reservoir engineers in many fields. Knowing if it exists a large preferred flow path can be really useful in order to optimize hydro-thermal stimulation. If the preferred flow path is able to transport to much fluid, projects such as CO2 or radioactive waste storage in porous tanks surrounded by cap rocks can be aborted. Thus collaboration with geologists should be obvious in order to find such fractures to evaluate their transmissivity property.