Abstract: Based on the daily precipitation and maximum temperature data of warm season (May to September) from 1979 to 2021 in Hunan Province, the extreme high temperature, extreme precipitation and no precipitation events are extracted by percentile threshold method. The high temperature composite situation in the first three days of extreme precipitation and no precipitation is discussed and classified into warm wet events and warm dry events. The temporal and spatial characteristics of the two composite events and the dependent structure characteristics between extreme climate factors are analyzed by combined use of EOF and copula function. The results show that: (1) spatially, composite events have occurred at all observation stations in Hunan Province, with the concurrence of warm dry events 20 times bigger than that of warm wet events. For warm and humid events, 11.35% of extreme precipitation events will experience extreme high temperature within the previous three days, with an average of 0.74 times per year, which were mainly distributed in the northwest mountains and Nanling Mountains. For warm dry events, 25.81% of the extreme high temperature events occurred within three days before no precipitation occurred, which were mainly distributed in the Dongting Lake Plain, eastern Hunan, and the valley area between Xuefeng mountain and Wuling Mountain. In terms of time, both types of composite events show an increasing trend and are concentrated in June and July. Especially since 2010s, the fastest growth rate mostly occurs in densely populated urban agglomerations. (2) The spatio-temporal anomaly characteristics of the two types of composite events show that the spatial distribution of the first mode of warm and humid events is consistent in the whole field, and the spatial distribution of the second mode shows that the central Hunan and Northern Hunan plain and valley areas are reversed. The first mode of the occurrence frequency of warm dry events shows that most parts of Hunan and southern Hunan have an inverse change, while the second mode shows that the eastern part of Hunan and the western mountain have an inverse change. (3) There is an obvious correlation between the maximum temperature and precipitation in Hunan Province, and the dependence structure results of t-copula function verify the specific correlation: from the joint probability distribution, the two on the whole show a negative correlation, while from the t-copula probability density function diagram of the two, they have an obvious positive correlation in the extreme values.