Cave Automatic Virtual Environment (CAVE) is a virtual reality (VR) environment that has not been fully studied due to its high cost and complexity in system integration. Previous CAVE-related studies mainly focused on comparing its effectiveness with other learning media, such as textbooks, desktop VR, or head-mounted display (HMD) VR. In this study, through the utilization of CAVE in a meteorology class, we concentrated on CAVE itself, measured how CAVE impacted learners' learning outcomes before and after using CAVE in an actual ongoing undergraduate-level class, and investigated how learners perceived their learning experiences. Quantitative data were collected to examine the students' knowledge acquisition and learning experience. We also triangulated the quantitative results with qualitative data from the interviews regarding learners' perceptions of the CAVE-enabled class and their knowledge mastery. The results indicated that their learning outcomes increased through learning with CAVE and that their perceptions of immersion, presence, and engagement significantly correlated with each other. The interview results showed a great fondness of and satisfaction with the learning experience, group collaboration, and effectiveness of the CAVE-enabled class from the learners. We also learned that the learners' learning experiences in CAVE could be further improved if we provided them with more learner-environment interaction, offered them a better sense of immersion, and reduced cybersickness. Implications of these findings are discussed.