As new techniques exploiting the Earth's ambient seismic noise field are developed and applied, such as for the observation of temporal changes in seismic velocity structure, it is crucial to quantify the precision with which wave-type measurements can be made. This work uses array data at the Homestake mine in Lead, South Dakota, and an array at Sweetwater, Texas, to consider two aspects that control this precision: the types of seismic wave contributing to the ambient noise field at microseism frequencies and the effect of array geometry. Both are quantified using measurements of wavefield coherence between stations in combination with Wiener filters. We find a strong seasonal change between body-wave and surface-wave content. Regarding the inclusion of underground stations, we quantify the lower limit to which the ambient noise field can be characterized and reproduced; the applications of the Wiener filters are about 4 times more successful in reproducing ambient noise waveforms when underground stations are included in the array, resulting in predictions of seismic time series with less than a 1% residual, and are ultimately limited by the geometry and aperture of the array, as well as by temporal variations in the seismic field. We discuss the implications of these results for the geophysics community performing ambient seismic noise studies, as well as for the cancellation of seismic Newtonian gravity noise in ground-based, sub-Hertz, gravitational-wave detectors.