探索性数据分析在西藏气候变化趋势研究中的应用

为了减少小样本时间序列资料中个别异常值对气候变化趋势分析的破坏性影响，采用探索性数据分析方法之一的三组耐抗线对西藏近30年气温、云量资料进行线性趋势分析，研究了西藏地区气温和云量的变化趋势及其区域分布，并讨论了二者的相关特征。通过与传统的最小二乘法相比，表明三组耐抗线用于西藏气候变化趋势分析具有明显优势，而用最小二乘法得到的西藏气候变暖估计偏高。分析认为，西藏气候变暖主要由最低气温增高显著所致，最低气温变暖趋势在春、夏季较明显，冬季相对不明显；西藏各地增温幅度在空间分布上不一致，中、西部地区呈较强的增温趋势，东部相对较弱；西藏高原总云量和低云量总体呈减少趋势，总云量和低云量变少分别与西部和中部地区的升温之间存在显著负相关关系。

APPLICATION OF EXPLORATORY ANALYSIS ON CLIMATE CHANGE IN TIBET

Based on three group resistance line(TGRL),and with the aim to decrease the error of the calculation of climate change,which can be significantly influenced by outliers,the variation trends of surface air temperature,extreme temperature and cloudiness in Tibet area from 1971 to 2000 were investigated,and the correlation between temperature and cloudiness change of the area was also analyzed.Compared with the results obtained by traditional least square regression method,the calculations of this study showed that the variation trends revealed by TGRL are less than by the latter,indicating that the traditional least square regression method might overestimate the degree of climate change in Tibet.Increasing trend of minimum temperature is the main factor causing the warming tendency in Tibet,and the warming in spring and summer is much more obvious than that in winter.There is a warming tendency with high(low) values in the west and inner(east) region,where the pattern of tendency distribution shows that localization is affected by the complex terrain.Both of the total and low level cloudiness in the area were decreasing in the past 30 years.It was also shown that variations of temperature have a strong correlation with cloudiness,and decrease in cloudiness is an important reason for the warming in Tibet.