In this paper, a method of phase space in modeling the climatological serie
s is put forward on the correlation dimension, the theory of phase space an
d the direct retrieval technique. Than time series of monthly mean temperat
ure are used to build dynamic models of regional climate in phase space in
which some dynamic characteristics can be discovered. The results illustrat
e that the methodology holds promise for a wide variety of applications. Fu
rthermore, the evolution characteristic and the interior interaction mechan
ism of the regional climate of southern China are also discussed.