Abstract: A large number of raindrop spectrum precipitation phenomenon instruments have been installed in China. With more and more equipment being installed, national measurement of precipitation and particle spectrum has become increasingly possible. However, comparison analyses with available data indicated that the accuracy of precipitation type recognition was relatively poor. We also found that the meteorological elements changed with rainfall events. So that, a comprehensive method by combining with meteorological elements has been proposed to improve accuracy of precipitation type recognition. The theory was similar with arti fi cial precipitation phenomenon observation. The thresholds of meteorological elements have been put forward to correct the results of precipitation type recognition. Due to the limitation of the working principle of the instrument measurement, the accuracy of the observation of a single instrument under the existing conditions cannot fully reach the arti fi cial observation level. With this regard, big data analysis method can be useful to solve this problem.