We have developed two methods to determine the tropospheric path delay for microwaves. One uses a dosed form model of the atmosphere in which two parameters are used to describe the decrease in temperature with height and the relation between total pressure and water vapor partial pressure, respectively. The other method uses numerical integration of refractivity profiles, which are computed from temperature and humidity profiles obtained by statistical regression. The two methods have been tested on two sets of observations and compared with simultaneous radio soundings. The best results show an improvement of 10% in one case and 30% in the other compared with a presently used model. Possibilities to improve the statistical model further are discussed. values for the constants k•, k 2 and k3 used by Saasta-file and the surface-correlated profile as estimates of moinen, which differ slightly from the values given by the water vapor profile. It can be seen that the im-Thayer [1974]. provement achieved by surface correlation is large J. W. Marini (internal NASA memorandum, 1974) for lower altitudes and that it becomes gradually modified (27)to smaller with increasing altitude. For further details about minimum-variance estimation see Melsa and AL = [1/f(tp, H)] Cohn [1978]. This method to obtain estimates of temperature (28) and humidity profiles has earlier been used to prosin (EL) + [B/(A + B)]/[sin (EL) + 0.015] vide the initial guess needed in atmospheric profiling ASKNE AND NORDIUS: TROPOSPHERIC DELAY FOR MICROWAVES 383

No abstract