关键词: 气候变化, 小波分析, 人工神经网络, 西南喀斯特地区

Abstract: Climate change has seasonal and regional characteristics. Many researchers have paid more attention to regional climate change in China. This paper analyzes characteristics and tendencies of climate change in the southwestern karst region of China in the recent 45 years (1960—2004) using wavelet analysis, artificial neural network and GIS spatial interpolation analysis, and so on. The results show that this regional climate change demonstrates significant periodic variations. The principal period of temperature variations is about 7 years, and the principal period of rainfall variations is about 25 years. It also has obvious seasonal characteristics and regional variations. By calculating seasonal temperature and rainfall variation rates with spatial interpolation, regional distributions of seasonal temperature and rainfall variations are obtained. Temperature and rainfall variations during the past 45 years are classified by means of SOFM network. Spatial distributions of the temperature and rainfall variations are obtained using spatial interpolation analysis. Finally, climate tendency is predicted with the RBFN model. Results show that variations of temperature and rainfall are both in a relatively low value period during 2005—2010.

Key words: climate change, wavelet analysis, artificial neural network, southwestern karst region, China