Landscape evolution provides insight into the tectonic and erosional processes that have shaped the topography observed today. Landscape evolution can be constrained using data such as thermochronometry, which records the amount of erosion through time, or using numerical landscape evolution models, which stimulate different tectonic or climatic scenarios. In both cases an idea of what the topography used to look like is required. Here we introduce a method to reconstruct fluvial topography based on the stream power model. The approach is based on an analytical solution to the steady state stream power model in which a single elevation within the drainage network is a function of the integrated channel steepness and χ values. The branching structure of a drainage network provides redundant information that can be exploited to infer spatial variations in channel steepness and a baselevel parameter. A single elevation pixel can be written as a sum of channel steepness multiplied by ∆χ values, and a set of elevation pixels can be combined as a system of equations. Solving this system of equations provides a map of channel steepness values which can be used to reconstruct topography. This approach is applied to examples from Greenland to infer a pre-glacial, or pre-incision, landscape that can be used to assess the climatic conditions required for ice growth. We also apply the approach to SE Tibet where an apparent relict landscape has been used to constrain geodynamic processes. We address the question that enables this surface to be used: how continuous was this surface ?