Here we use synthetic data to explore the performance of forward models and inverse methods for helioseismic holography. Specifically, this work presents the first comprehensive test of inverse modeling for flows using lateral-vantage (deep-focus) holography. We derive sensitivity functions in the Born approximation. We then use these sensitivity functions in a series of forward models and inversions of flows from a publicly available magnetohydrodynamic quiet-Sun simulation. The forward travel times computed using the kernels generally compare favorably with measurements obtained by applying holography, in a lateral-vantage configuration, on a 15-hour time series of artificial Dopplergrams extracted from the simulation. Inversions for the horizontal flow components are able to reproduce the flows in the upper 3Mm of the domain, but are compromised by noise at greater depths.

The subsurface structure of an "average" supergranule is derived from existing HMI pipeline time-distance data products and compared to the best helioseismic flow model detailed in Duvall and Hanasoge (2013). We find that significant differences exist between them. Unlike the shallow structure predicted by the model, the average HMI supergranule is very extended in depth, exhibiting horizontal outflow down to $7$--$10$~Mm, followed by a weak inflow reaching a depth of $\sim20$~Mm below the photosphere. The maximal velocities in the horizontal direction for the average supergranule are much smaller than the model, and its near-surface flow field RMS value is about an order of magnitude smaller than the often-quoted values of $\sim250-350$~$\rm{m\,s^{-1}}$ for supergranulation. Much of the overall HMI supergranule structure and its weak flow amplitudes can be explained by examining the HMI pipeline averaging kernels for the near-surface inversions, which are found to be very broad in depth, and nearly identical to one another in terms of sensitivity along the $z$-direction. We also show that forward-modeled travel times in the Born approximation using the model (derived from a ray theory approach) are inconsistent with measured travel times for an average supergranule at any distance. Our findings suggest systematic inaccuracies in the typical techniques used to study supergranulation, confirming some of the results in Duvall and Hanasoge (2013).

Linear time-distance helioseismic inversions are carried out using several filtering schemes to determine vector flow velocities within two $\sim100^2\,{\rm Mm^2}\times 20\,{\rm Mm}$ realistic magnetohydrodynamic sunspot simulations of 25~hr. One simulation domain contains a model of a full sunspot (i.e. one with both an umbra and penumbra), while the other contains a pore (i.e. a spot without a penumbra). The goal is to test current helioseismic methods using these state-of-the-art simulations of magnetic structures. We find that horizontal flow correlations between inversion and simulation flow maps are reasonably high ($\sim0.5$--0.8) in the upper 3~Mm at distances exceeding 25--30~Mm from spot center, but are substantially lower at smaller distances and larger depths. Inversions of forward-modeled travel times consistently outperform those of our measured travel times in terms of horizontal flow correlations, suggesting that our inability to recover flow structure near these active regions is largely due to the fact that we are unable to accurately measure travel times near strong magnetic features. In many cases the velocity amplitudes from the inversions underestimate those of the simulations by up to 50\%, possibly indicating nonlinearity of the forward problem. In every case, we find that our inversions are unable to recover the vertical flow structure of the simulations at any depth.

Linear time-distance helioseismic inversions are carried out for vector flow velocities using travel times measured from two $\sim 100^2\,{\rm Mm^2}\times 20\,{\rm Mm}$ realistic magnetohydrodynamic quiet-Sun simulations of about 20 hr. The goal is to test current seismic methods on these state-of-the-art simulations. Using recent three-dimensional inversion schemes, we find that inverted horizontal flow maps correlate well with the simulations in the upper $\sim 3$ Mm of the domains for several filtering schemes, including phase-speed, ridge, and combined phase-speed and ridge measurements. In several cases, however, the velocity amplitudes from the inversions severely underestimate those of the simulations, possibly indicating nonlinearity of the forward problem. We also find that, while near-surface inversions of the vertical velocites are best using phase-speed filters, in almost all other example cases these flows are irretrievable due to noise, suggesting a need for statistical averaging to obtain better inferences.