Several types of all-sky viewing Fabry-Perot Interferometers (FPI) have been developed since the 1990s for ground-based remote sensing of thermospheric winds. The Scanning Doppler Imager (SDI) is one such instrument, which provides temporally simultaneous line-of-sight observations from hundreds of independent look directions per instrument exposure. A geographically distributed network of such instruments increases spatial coverage and, at many locations, also provides overlapping observations along multiple independent lines-of-sight. Together, these characteristics significantly increase the density and fidelity that is possible for reconstructed thermospheric vector wind fields, compared to a traditional narrow-field FPI, but at the cost of complexity and difficulty.

Presently, we describe an application of inverse theory to reconstruct three-component vector thermospheric neutral wind fields using data from multiple SDI instruments. The salient features of the method used here are the ability to reconstruct three-component winds on a dense grid that is sampled regularly in latitude, longitude, and time, without assuming any a-priori underlying structure of the winds. This requires solving an inverse problem that does not in general yield a unique solution unless additional constraints are enforced. We describe this step, also known also as regularization, along with the strategy used to maximize the spatial resolution of the derived wind fields by automatically determining the minimum level of regularization that can produce stable inversions. We present example results obtained from applying this technique to one night of data from a network of SDIs in Alaska, and discuss the implications of these results for current understanding of thermospheric dynamics.