Interferometric synthetic aperture radar (InSAR) observations of ice-shelf flow contain ocean-tide and atmospheric-pressure signals. A model-based correction can be applied, but this method is limited by its dependency upon model accuracy, which in remote regions can be uncertain. Here we describe a method to determine two-dimensional ice-shelf flow vectors independently of model predictions of tide and atmospheric pressure, by stacking conventional and multiple aperture InSAR (MAI) observations of the Dotson Ice Shelf, West Antarctica. In this way we synthesize a longer observation period, which enhances long-period (flow) displacement signals, relative to rapidly varying (tide and atmospheric pressure) signals and noise. We estimate the error associated with each component of the velocity field to be ~22 ma-1, which could be further reduced if more images were available to stack. With the upcoming launch of several satellite missions, offering the prospect of regular short-repeat SAR acquisitions, this study demonstrates that stacking can improve estimates of ice-shelf flow velocity.