Optical interferometry is the only means of directly measuring the sizes of stars. The most precise angular diameter measurements, however, depend on measuring complex fringe visibilities V at spatial frequencies where Re(V ) crosses zero. We can then use the spatial frequency B ⊥ /λ 0 of the zero crossing as a measure of the stellar diameter via θ UD,0 ≈ 1.22λ 0 /B ⊥ , where λ 0 and is the wavelength at which Re(V ) = 0 when observed with a baseline length B ⊥ projected toward the star, and θ UD,0 is the equivalent uniform disk diameter. The variation in limb darkening with wavelength leads to a corresponding variation in θ UD,0 with λ, even at fixed B, which allows us to measure the limb darkening in detail and probe the structure of the atmosphere. However, in order to take meaningful data at those spatial frequencies, we need some form of bootstrapping, in wavelength, baseline length, or both. Reduction of these bootstrapped data benefits greatly from the increase in SNR offered by coherent averaging. We demonstrate the effect of limb darkening on θ UD,0 (λ) with simulated observations based on model atmospheres, and compare them to coherently averaged NPOI data.
THE ZERO-CROSSING IDEAThe majority of interferometric stellar diameter measurements are made at spatial frequencies u within the first null of the V 2 (θ, u) visibility curve. Within this range, the difference between the best-fit uniform disk curve and a curve that takes limb darkening into account is small. Therefore, the usual method of determining a diameter from these data is to fit a uniform-disk model to the observed visibilities, and multiply the resulting uniform-disk diameter θ UD by a correction factor appropriate to the stellar type to obtain the limb-darkened diameter θ LD .In principle, we could make more-detailed diameter measurements by fitting a uniform disk curve through each measured V 2 (u) point. If the data were from several wavelength channels on the same baseline, we would obtain a set of uniform disk diameters θ UD (λ) resulting from the individual uniform-disk V 2 (B ⊥ /λ) curves (B ⊥ is the baseline length projected toward the star). In practice, the visibility calibration is usually not accurate enough to give an interesting degree of precision for θ UD (λ): when V 2 ≈ 0.6, a 1% error in V 2 corresponds to a 1% error in θ UD (λ), while at V 2 ≈ 0.4, a 1% error in V 2 corresponds to a 0.5% error in θ UD (λ).However, we can make a much more precise determination of θ UD (λ) if we have data that straddles Re [V (u)] = 0 -for this discussion, it is more useful to look at Re(V ) than at V 2 -because any (multiplicative) calibration * tom.armstrong@nrl.navy.mil; phone 1 202 767 0669; fax 1 202 404 8894Optical and Infrared Interferometry III, edited by Françoise Delplancke, Jayadev K. Rajagopal, Fabien Malbet, Proc. of SPIE Vol. 8445, 84453K · © 2012 SPIE · CCC code: 0277-786/12/$18 ·