We present a feasibility study for a high-frame-rate Short-baseline auroral tomographic imaging system useful for estimating parametric variations in the precipitating electron number flux spectrum of dynamic auroral events. Of particular interest are auroral substorms, which are characterized by spatial variations of order 100 m and temporal variations of order 10 ms. These scales are thought to be produced by dispersive Alfvén waves in the near-Earth magnetosphere. The auroral tomography system characterized in this paper reconstructs the auroral volume emission rate, to estimate the characteristic energy and location in the direction perpendicular to the geomagnetic field of peak electron precipitation flux, using a distributed network of precisely synchronized ground-based cameras. As the observing baseline decreases, the tomographic inverse problem becomes highly ill-conditioned; as the sampling rate increases, the signal-to-noise ratio degrades and synchronization requirements become increasingly critical. Our approach to these challenges uses a physics-based auroral model to regularize the poorly observed vertical dimension. Specifically, the vertical dimension is expanded in a low-dimensional basis, consisting of eigenprofiles computed over the range of expected energies in the precipitating electron flux, while the horizontal dimension retains a standard orthogonal pixel basis. Simulation results show typical characteristic energy estimation error less than 30% for a 3-km baseline achievable within the confines of the Poker Flat Research Range, using GPS-synchronized electron-multiplying charge-coupled device cameras with broadband BG3 optical filters that pass prompt auroral emissions.