Tracking the movement and electrical behavior of weakly electric fish offers insight into the interplay of sensing and social dynamics in the wild, along with an understanding of broader changes in the fishes’ habitats. Localizing these fish using their electrical signals is challenging due to the nonlinear mapping between a fish’s pose and the voltage it generates at any point in space, as well as signal interference among groups of fish. We present an algorithm to prune the space of possible fish locations given measurements from an electrode array, approximating the fish as a planar oscillating dipole. The algorithm relies on interpolation of the potential field and its gradient between neighboring electrodes. We then use this algorithm to initialize probabilistic trackers–a particle filter and an extended Kalman filter–and show that under conditions where the dipole model holds, the algorithm allows us to track the fish more efficiently than if we assume no prior knowledge of the fish’s pose. We expect that when combined with probabilistic data association methods, our method will be useful in associating signals to fish during periods of interference. We also hypothesize that our method could be used for exact localization, given an interpolator that reflects the electric field characteristics of a planar dipole. Exploring data association methods and novel interpolation schemes will be the next step in our research.