A model of the latitudinal distribution of the solar wind and its time variations during recent decades is a starting point in many areas of heliosphere research. In-situ measurements of the solar wind parameters have been performed since the beginning of the space age, but they are limited to a narrow band around the solar equator. Although Ulysses performed in-situ measurements of solar wind parameters at high heliolatitudes, the continuous in-situ measurements at high heliolatitudes is lacking. In order to fill the gap in the coverage of solar wind speed and density at high heliolatitudes, computer-assisted tomography of interplanetary scintillation observations (IPS) is performed. The solar wind speeds at all heliolatitudes are provided in the form of maps Carrington-period maps. Their general fidelity was demonstrated by comparison of the IPS speed profiles at high heliolatitudes with those from Ulysses observations. However, because of operational reasons, the Carrington speed maps have insufficient coverage during the year, and sometimes feature extreme unrealistic values, which contribute to undesirable background. During determination of the average solar wind speed profile, the background points accumulate and affect the obtained results. We introduce a novel method of the IPS data processing to reduce the impact of the undesirable background. Using this method, we develop a new model of the time- and heliolatitude evolution of the solar wind speed and density based on Legendre functions. We demonstrate that the coefficients of the Legendre functions are correlated with carefully selected solar proxies. This enables reconstructing the latitudinal profiles of the solar wind speed and density beyond the period of IPS data (i.e. since 1985 to present), both backwards and forwards in time. In effect, we present a new phenomenological solar wind model which brings to us a unique possibility to fill the gap in the solar wind 3D structure before 1985, as well as in the present day before the IPS data are available. We present the model formulation, its overall properties, and its limitations. Example estimations of the SW 3D structure before IPS-era and their implications will also be shown.