Abstract: Spacecraft to be launched at sea are exposed to harsh, hot, and humid corrosive environments for extended periods. These regional climatic conditions significantly constrain their operational performance and storage reliability. However, the correlation between test profiles derived from existing environmental testing methods and actual environmental data remains insufficient. In this study, the coastal area of Wanning (in Hainan province, China) was selected as the research site. A distributed monitoring system was used to collect real-time temperature and humidity data from three types of sites: outdoor exposed areas, sheds, and storage rooms. The K-means clustering and K2 Bayesian network structure learning algorithms were applied to characterize the intrinsic structures of the temperature and humidity data and to explore their underlying patterns. The results showed that the temperature and humidity distributions could be clustered into three typical patterns: moderate temperature with high humidity, high temperature with low humidity, and low temperature with low humidity. A Bayesian network model was established to reveal the probabilistic relationships among months, sites, and temperature–humidity patterns, enabling the generation of probabilistic environmental profiles based on actual mission parameters. The proposed method demonstrates excellent adaptability in identifying coastal environmental characteristics and offers a viable methodology for the design of spacecraft adaptability to the hot and humid marine environment, and the customized design of test conditions.