The exploitation of multifrequency differential attenuation measurements at microwaves made between two LEO
satellites in limb mode is the ground of the NDSA (Normalized Differential Spectral Attenuation) approach for
estimating integrated tropospheric water vapor profiles through multifrequency measurements at 17, 19, 21, 179 and 182
GHz, plus 32 GHz for liquid water detection and correction (whenever possible). Such measurements are affected by
two kinds of impairments, the first generated by thermal noise at the receiver, the second generated by the signals’
fluctuations due to the variations of the tropospheric refraction index and referred to as scintillation disturbance.
Characterizing scintillation for simulating its effects to evaluate NDSA performance is not easy in general: in particular,
it is quite hard (and also rather questionable so some extent) to relate the scintillation parameters to a given simulated
atmospheric situation. For this reason, in the past years we limited ourselves to evaluate the NDSA performance by
accounting for scintillation in a parametric way, independently of the atmospheric context in which simulations were
carried out. In this paper, instead, we show the first results of the NDSA performance analysis based on a completely
different approach, where the scintillation profiles and parameters are directly derived from the simulated atmospheric
context, based on a procedure that starts from high resolution radiosonde data. A brief critical analysis of such an
approach is proposed, evidencing some aspects related to the current knowledge of the scintillation spectra and
parameters. The NDSA performance analysis based on certain hypotheses for the scintillation characteristics is then
shown for some selected simulated atmospheric conditions.