Abstract: In this paper, monthly mean temperatures of 1873-1980 for Shanghai and Guangzhou are used to extend the 1-D climatic time-series into a multi-dimensional phase space. Calculation results show that fractional dimensions attractors develop in the phase space in the short-range climate evolution displayed by the monthly mean temperatures over the summer monsoon region in China. Their dimensions are d1=3.4 and d2=2.3 respectively.It is seen that the climate evolution would be best described by four variables or a dynamic model with no less than four steps, required by the development of these chaotic attractors in the multi-dimensional phase space.