A Simple Method of Correcting Magnitudes for the Errors Introduced by Atmospheric Refraction

We show that the errors due to atmospheric refraction are present in the magnitudes determined with the Difference Images Analysis method. In case of single, unblended stars the size of the effect agrees with the theoretical prediction. But when the blending is strong, what is quite common in a dense field, then the effect of atmospheric refraction can be strongly amplified to the extent that some cases of apparently variable stars with largest amplitudes of variations are solely due to refraction. We present a simple method of correcting for this kind of errors. It is generally known, that atmospheric refraction is a source of serious problems when one is doing astrometry (Evans and Irvin 1995 and references therein). The refraction shifts positions of observed stars toward the zenith, and as it is color dependent the resulting shifts of stars of different colors are different. The photometry is much less affected by refraction. Usually, a measurement of stellar magnitude is performed by three parameter fit of the stellar image to a known point spread function. Beside the magnitude that is the main objective of the measurement, two angular coordinates are also determined. An accompanying determination of the image position absorbs the effect of atmospheric refraction. The image position is shifted but the resulting value of magnitude is unaffected. In fact it is almost unaffected. The refraction causes the point spread function to be elongated in the direction toward the zenith. When the field size is large then the zenith distance varies and the degree of the PSF elongation is changing when moving across the field. This can be handled by using any position dependent point spread function technique what in the case of large fields is usually necessary for other reasons anyway. There could also be some tiny dependence of the point spread function itself on spectral energy distribution of measured objects. An unresolved binary composed of stars with much different spectral types should have the point spread function more elongated in the direction to the zenith compared to the normal case of a single star. But such an effect of point spread function distortion should be