Abstract: Sea surface temperature is an important disaster-promoting environmental factor for nearshore marine heatwave events, and revealing its variation pattern will help predict the occurrence of future heatwave events. In this paper, the association rules used to express the changes of sea surface temperature before and during heat wave events are defined as event patterns, and a pattern discovery method based on restricted motif association rule mining was proposed. The STAMP algorithm was used to mine the motifs, and the MDL scoring strategy was used to segment the motifs to form candidate association rules. Combined with the constraints of the occurrence of heat wave events, the pattern of marine heat wave events was extracted. In this paper, the sea surface temperature data from three stations in China's coastal waters are used to conduct a marine heat wave event pattern discovery experiment. The results show that the marine heat wave in China's coastal waters is seasonal, mainly occurs in summer, and lasts for more than 20 days. The marine heat wave event pattern has a regular heating and falling trend, the heating rate and cooling rate are symmetrical, and before the heat wave occurs, the sea surface temperature has a short heating interval.