The identification of supercooled water in convective clouds has always been a difficult point in meteorological sounding. Based on Doppler spectra of a Ka-band millimeter-wave radar and relevant radiosonde data, an algorithm for identifying and retrieving supercooled water in convective clouds of the Tibetan Plateau was proposed. Subsequently, retrieval effects of the algorithm were analyzed using two convective cases, and verified by comparing with measurements of a co-located microwave radiometer (MWR). Finally, the difference in results of the algorithm and other three previous methods was also discussed. The main findings are as follows: the stratocumulus, cumulus congestus, and altocumulus clouds over Nagqu are dominated by updrafts with rapid changes on the hydrometeor phase in the vertical orientation, resulting in widely distributions of the formed supercooled particles in terms of their reflectivity, effective radius and liquid water content. Supercooled particles of different convective cloud types also locate at different cloud body positions. The velocity of the in-cloud updraft is highly and positively correlated with the reflectivity, effective radius and liquid water content of supercooled water. They possess similar temporal variations and coincident spatial distributions. The radar-derived liquid water path is also proved to agree well with the counterparts of MWR with similar temporal variations and value peaks. Their correlation coefficients can approach 0.63~0.79. Compare with three previous methods, results from the proposed algorithm can be more reasonable on the retrieved supercooled water positions and parameters.