This contribution wants to explore the potentials of informing the design of large-scale networks of paths with predictive models, capable of combining a simulation of users' behaviors within the space and the geo-morphologic intrinsic characteristics of the site. Our work investigates how a deeper knowledge of topology allowed by new technologies could support the application of behavioral modeling at the landscape scale, indirectly informing the creation of paths in the absence of reliable in situ observations on the actual use of a determined area. Thus, the creation of a predictive model capable of simulating users' behaviors according to pre-established parameters can be developed thanks to the cross use of parametric design softwares and territorial informative systems, with the aim of managing and processing a series of useful information deduced both from experience and from the established outputs of studying desire lines, for defining a bundle of potential routes in a given spatial context.