A new technique for estimating the global-scale pattern of magnetospheric convection in the ionosphere is presented. The technique uses the SuperDARN line-of-sight velocity observations combined with an empirical convection model and the assumption that the resulting velocity field is divergence free. In contrast to other techniques for convection estimation, it does not express the velocity field in terms of known basis vectors and it does not assume that the velocity can be determined from a static potential. The velocity is estimated by applying Bayesian inverse theory to the input data, model, and constraints. Linear equations for the plasma velocity at every point in the domain are solved simultaneously in a least-squares sense. Application of the technique results in convection patterns with spatial resolution equal to the calculation grid. The resulting patterns conform with expectations based on the observed IMF conditions and display features that show close correspondence to simultaneously observed features in the auroral luminosity.

A new model of high-latitude convection derived using machine learning (ML) is presented. The ML algorithm random forests regression was applied to a database of velocity observations from the Super Dual Auroral Radar Network (SuperDARN). The features used to train the model were the IMF components Bx, By, and Bz; the solar wind velocity, vsw; the auroral indicies, Au and Al; and the geomagnetic index, SYM-H. The SuperDARN velocities were separated into north-south, and east-west components and sorted into a magnetic local time - magnetic latitude grid that ran from 55° to the magnetic pole with a bin size of 2° in latitude, and 1-hour in MLT. Separate models were created for each velocity component in each bin of the grid. It is found that even though the models in each bin are independent of one another a coherent convection pattern is formed when the models are viewed in aggregate. The resulting convection pattern responds to changes in the auroral indicies by expanding and contracting in a way that is consistent with expectations for a substorm cycle. Further it is found that the mean-squared difference between predictions of the model and observed values of the velocity are substantially lower than the same quantity calculated for an existing climatology that was not formed with ML techniques