From late December 2022 to mid-January 2023, California experienced a series of nine consecutive atmospheric rivers (ARs) making landfall. These intense weather events led to devastating consequences, including severe flooding and mudslides, widespread power outages, and at least 22 fatalities. However, limited research has focused on the spatiotemporally compounding ARs. In particular, linkages among the variation of the intervals between AR landfalls, different land responses, and the large-scale circulation remain undocumented. Utilizing an unsupervised machine learning technique, we define an AR cluster as a short period that consists of back-to-back ARs. Here we show that the characteristics and impacts vary significantly by cluster density, which is the fraction of AR conditions within a cluster. Focusing on the landfalling ARs over the U.S. West Coast, we found that clusters with high density (such as the 2023 series of nine ARs in three weeks) consist of higher AR categories and higher likelihood for extreme precipitation and severe land surface response. By using reanalysis and model simulation, we showed the key circulation pattern for AR clusters is mainly attributed to subseasonal variability. In addition, the third mode of geopotential height variability modulates the occurrence and density of AR clusters. Furthermore, we demonstrated that AR clusters with higher density and category will be more frequent in warming climates. Our study highlights the important role of AR clusters in the planning and development of climate adaptation and resilience.