Uncertainty: How Error Sources Affect the Stitching
Because it can take advantage of much redundant information among overlapping subapertures, the stitching algorithm can separate or depress measurement errors from various sources. Stitching interferometry is even claimed to have lower uncertainty than individual measurements because systematic errors are self-cali- brated.62 However, most people may doubt it simply based on empirical observations. Their worry comes first from the experience that subaperture measurements are more susceptible to noise due to vibrations, air turbulence, or thermal instability. Taking subaperture measurements one by one is time consuming and requires frequently positioning and nulling with a multiaxis mechanical stage. How do the subaperture noises affect the stitching? This is the first question. In addition, coupling of noises, systematic errors, misalignment-induced aberrations, and surface error may deactivate the power of stitching optimization. For example, the surface error of pure spherical aberration on the full aperture results in identical subaperture aberrations (with the same amount of off-axis distance) that are mostly composed of astigmatism and coma. Here, “identical” means the aberrations appear the same as systematic errors that will be self-calibrated out from the stitching result of surface error. How can we decouple or identify them to produce a real stitched surface error map? These questions remain unanswered.