The invention discloses a thunderstorm clustering method and device based on massive thunder and lightning data, a computer device and a storage medium. The method comprises the following steps: performing initialization processing on all lightning data in any region in any given time period, setting characteristic parameters of thunderstorm clustering, performing time scale pre-clustering on a lightning data set to obtain all data subsets, and selecting a clustering method according to the sizes of the data subsets; using a conventional clustering algorithm when the data size is small, and executing the step 4 when the data size is large; partitioning the data subsets, and synchronously clustering the partitions; clustering nodes in a partition boundary region, and finding out a clustering result relationship between adjacent partitions; combining partition clustering results; writing a clustering result into files. According to the method, the time attribute of the thunder and lightning data is converted into the space attribute, so that the clustering speed is higher and is more than three times faster than that of a conventional DBScan algorithm, and the larger the data size is, the more obvious the effect is. Meanwhile, the time and space attributes of the lightning data are considered, so that the accuracy of a clustering result is improved.