Proceeding from the present-day and future needs of astrometry, current researches into astronomical refraction as well as its prospective development are analysed. Particular attention is dedicated to the constructing of new refraction tables, determination of the refractional influences by way of experiments, site selection of the observing stations and to the preventive measures to be taken at them. Need and usefullness are emphasized of the cooperation of experts specialized in astronomical, terrestrial, photogrammetric, radio and other kinds of refractional influences.

In the last few years there was no paper in the astronomical literature reporting on new measurements of radioastronomical refraction. But there are some important review papers on different aspects of this topic, e.g. on the ionosphere by M.M. Komesaroff (1960) and T. Hagfors (1976), on the prediction of tropospheric refraction with ground based meteorological data by B.R. Bean (1962), B.R. Bean and G. Teleki (1974), and R.K. Crane (1976), on range measurements, predicted again by surface weather data, by H.S. Hopfield (1971).

For astrometry only interferometric observations give the necessary positional accuracy. The result of the interferometer is not affected, if the delay by the atmosphere is the same for each antenna, therefore this technique is not sensitive to normal refraction.

Differential delays can originate in the uneven distribution of water vapor in the atmosphere. Wesseling et al (1974) measured these differential delays with an infrared hygrometer; their attempts to correlate the water vapor with the observed interferometer phase were only partially successful.

Hinder (1970) has investigated the differential delays as function of baseline and of season. One can conclude that improvements of positional accuracy either by going to longer baselines or to shorter wavelengths depend on the success to predict the differential delays either by infrared hygrometers or similar means.

The role of urban heat islands in producing systematic isopycnic tilts is explored in more detail, and with greater rigor, than in Part I of this series. (Perth, 1974).

Specifically, a three dimensional integration is carried out, and light rays are, in effect, “traced” through the resulting perturbation field by evaluating the integral of anomalous refraction. This is done for various values of the parameters, viz., wind direction and observatory location relative to the heat island, strength of the central perturbation, zenith distance of the observed object, etc.

It is stressed that heat islands are not the only source of such systematic effects.

Finally, a brief discussion of some possible methods of determining observationally the effects here treated theoretically, as well as other site dependent effects, is appended.

There exists an empirical relation between the anomalous refraction and the atmospheric density in the surface layer. From the relations the variations of scale height for each night can be determined by the temperature and pressure in the surface layer. A correction term to the refraction table is derived in an analytical expression.

The existing refraction theories for a one-dimensional model atmosphere do not account for anomalous refraction. The latter can be calculated using a three-dimensional model atmosphere.

The astrometric plate reduction needs the knowledge of formulae expressing the transformation of the configuration of stellar positions on the celestial sphere into the corresponding observed figure of the photographic images on the plate. This transformation includes phenomena such as the atmospheric refraction, the optical distortion, the plate position on the instrument, etc. of which the relative importance may vary from one case to another and may lead thus to various formulae.

It is well known that the turbulent state of the atmosphere gives rise to fluctuations of the coefficient of light refraction. In astrometric observations these fluctuations produce the phenomenon of image motion which serves as the source of accidental errors of observations. The need for the study of the effect of atmospheric turbulence is documented in (G.Teleki, 1967) and others.

Approximations have been removed from the derivation of the coefficients in the binomial expansion for astronomical refraction, , allowing the calculation of any number of terms to any precision desired. The range of the refraction formula has been extended to greater zenith distances (<90°) by inserting a damping factor into the binomial formula, truncating the expansion at a proper point, and rearranging the terms. Another, computer-manipulated series has been developed for zenith distances at or near the horizon. Further applications include the calculation of photogrammetric and parallactic refractions, as well as range corrections in satellite geodesy.

A continued fraction was derived for the summation of the asymptotic expansion of astronomical refraction. Using simple approximations for the last denominator of the fraction, accurate formulae, useful down to the horizon, were obtained. The method is not restricted to any model of the atmosphere and can thus be used in calculations based on actual aerological measurements.

The astronomical refraction mainly depends on the vertical structure or the height profile of atmospheric density. Concerning the vertical structure of atmospheric density, various hypotheses were presented during the last century. Among them Newton's hypothesis of equal temperature and Ivory's one that temperature diminishes at a uniform rate with height appear to represent the height profile of temperature approximately in the lower part of the stratosphere and in the troposphere respectively.

The formulation to calculate the precise refraction by means of the primitive equations by an electronic computer was made to be used for the calculation of the atmospheric refraction in the astrometry and the satellite geodesy with upper air data obtained with the atmospheric soundings.

In 1962 a photographic vertical circle (PVC) for determining absolute declinations was designed and constructed at the Pulkovo Observatory under the supervision of Prof. M.S. Zverev (Zverev, M.S., 1960). The Maksutov meniscus cassegrain system (D = 200 mm, f′ = 2000 mm) was used for the first time in astrometry. Its construction is symmetric and it is automatically operated. In 1963–1966 the declination observation of bright and faint stars of the southern hemisphere was made with this instrument (Zverev, M.S. et al., 1966). At present after the reconstruction the PVC has been used again for determination of absolute declinations (Bagildinsky, B.K., Usanov, D.S., Smirnov, B.N., 1978).

Atmospheric refraction, with its non-predictable variations, is one of the dominant sources of error in the precise determination of the apparent position of a star. It appears that this will be the dominant error source for the new generation of instrumentation now being installed for the determination of stellar catalog and Universal Time (i.e., the USNO 65-cm PZT). The effects of anomalous refraction measurements of stellar position propagate to the various quantities derived from this data. This includes the compilation of stellar catalogs, the determination of Universal Time, and variations in the latitude.