Abstract: Using the adaptive multiresolution data filter established by paper,this paper studies the multihierarchical structure and jump features of climate changes for the Globe, China and Yunnan during the last one hundred years. The results show that (1) the main features of the climatic variations for the Globe during the last one hundred years are warming and they can be described as 3 hierarchies corresponding to the larger-scale changes. They are the cold period before 1919, the warm period from 1920 to 1978 and the warmer period after 1979. The larger-timescale's hierarchical changes obviously appear features of climate jump. The years 1920 and 1979 are the jump points of climate change. The climatic variations for the Globe during the last 100 years appear to have more multihierarchical structure and more jump features corresponding to the smaller-scale changes. The filtered data of multiresolution not only conserve hierarchical structure and jump points of the larger scale changes, but also increase the new hierarchical structure and the new jump points. (2)The surface air temperature variations for China and Yunnan province are analogous but are not all analogous to the Globe. The main contrasts are that the Globe is the weaker cold period from 1955 to 1978 and the year 1979 is the jump point of warming,while China and Yunnan province are the stronger cold period from 1955 to 1986 and the year 1987 is the jump point of warming. (3) The hierarchical structure and the jump features of surface air temperature variations for Yunnan province in 4 seasons during the last 100 years are roughly analogous but of difference. The contrasts are that the Winter and the Spring are analogous to China, but the Summer and the Autumn are analogous to the Northern Hemisphere and the Globe.