Abstract. We consider waves propagating through multiscale media. Much is known about waves propagating through a medium that satisfies a scale separation assumption with random fluctuations on a microscale. Here we go beyond this situation and consider waves propagating through a medium defined in terms of a long range process. Such a medium can, for instance, be modeled in terms of a one-dimensional fractional Brownian motion with variations on a continuum of scales. Fractal medium models are used to model, for example, the heterogeneous earth and the turbulent atmosphere. We set forth a framework using the theory of rough paths in which propagation problems of this nature can be analyzed in the case with anticipative medium fluctuations with a Hurst exponent H > 1/2. We show how the wave interacts with the medium fluctuations in this case and that the interaction is qualitatively different from the situation where the medium satisfies a separation of scales assumption. In the long range case considered here the travel time depends strongly on the particular medium realization, but in fact the pulse shape does not.Key words. wave propagation, random media, long range processes, fractional Brownian motion AMS subject classifications. 34F05, 34E10, 37H10, 60H20 DOI. 10.1137/07068610X1. Introduction. Modeling in terms of a multiscale medium is important for propagation problems in, for instance, the earth's crust, the turbulent atmosphere, turbulent boundary layers, sea ice, and outer space [9,15,19,20,34,42,44]. Communication, remote sensing, and laser beam propagation schemes are affected by bad weather and multiscale medium variations. Large scale research projects (the ABLE ACE program Kirtland AFB, for instance [43]) have focused on gathering atmospheric turbulence data and numerically simulating propagation of wave fields through synthetic turbulence models that derive from these. Above a boundary layer atmospheric turbulence may occur within a stratified environment, and the turbulent temperature variations may be highly anisotropic; see [16,36]. Rough and long range medium fluctuations associated with multiscale modeling are also important in medical imaging, device modeling, and nuclear technology, to name a few. The zone in between different tissue types (or in between different dielectrica) may in particular be strongly heterogeneous with variation on a continuum of scales. In general, the detailed pointwise variation of a multiscale medium, the refractive index, say, cannot be identified. However, the statistics of this variation can be characterized. Optimal design of, for instance, imaging and communication algorithms requires insight about how the wave is affected by the rough medium fluctuations, that is, the nonlinear coupling between medium and wave field statistics. This is particularly the case with modern high resolution sampling and imaging technology. Insight about the wave medium interaction is also important in a range of other applications like design of sound-absorbing materials and non...