This paper presents a series of improvements made in the Topside Sounders Model assisted by Digisonde (TaD), verification results on these improvements, and its validation at the topside part of the profile. The TaD is based on the simple empirical functions for the O + /H + transition height (h T ), the topside electron density scale height (H T ), and their ratio, based on the Alouette/ISIS database. From its first release, published some years ago, TaD offers analytical formulas for obtaining the shape of the vertical plasma distribution in the topside ionosphere and plasmasphere. This first version of the TaD model (TaDv1) is using Digisonde measured parameters of the F layer maximum density, its height, and its scale height to specify the profiler's characteristics at its lower boundary. TaDv1 models separately the O + and H + density profiles, providing the H + scale height in the plasmasphere, extracted from ISIS-1 topside sounder data, as a function of geomagnetic latitude. The upgraded version of TaD (TaDv2) provides calculation of O + , H + , and He + density distributions in transition region between topside F region and plasmasphere, extracted from the analysis of the electron density profiles from ISIS-1, and in addition approximates the plasmaspheric scale height as a function of altitude, latitude, local time, and season using an optimization procedure to achieve best fit with the measured profiles. These improvements, which concern the part of the profile above the transition height, are presented in detail in the first part of the paper. In the second part we present statistical results for the verification of the model's improvements that show that the optimization procedure contributes to a reduction of the model error of more than two times. The model validation for the topside part of the profile is presented in the last part of this paper, comparing Incoherent Scatter Radar (ISR) electron density profiles (EDP) with the model reconstructed profiles. Comparison with measured EDP from ISR at middle latitudes gives a model error of 3TECU which is close to the GNSS measurement error. Further improvements of TaD reconstruction model are attempted in a followon paper, mainly targeted to the reliable operational implementation of the algorithm.