This paper describes an innovative approach to radio channel characterization in UMTS and LTE mobile networks. In place of traditional drive tests (DT), which employ a single test mobile, a massive collection of georeferenced radio measurements is made from a wide population of user equipment (UE). This is possible with new 3GPP features, called ''minimization of DTs'' (MDT), which are implemented in the last generation UEs and enable the reporting of additional periodical measurements, including GPS position and estimated UE distance (i.e. delay) over the radio path. This opens to new fields of investigation in the mobile radio channel, unreachable with the legacy DT approach, such as multipath and Doppler analysis. The UE MDT data of UMTS and LTE RAN of Telecom Italia Mobile, in the Italian midsized city of Bologna, have been statistically analyzed. The big data elaboration has been performed with the Nokia proprietary system ''GeoSynthesis.'' The results give a high-resolution geographical view of the abovementioned channel phenomena affecting the quality and user experience. They are in good accordance with network performance indicators: the higher the multipath time, the worse the decoding performance of radio blocks (block error rate). The estimated Doppler shift also fits the known mobility patterns in the urban environment.