全球变暖是从二十世纪五十年代开始的全球平均气温上升现象.希望通过对不同时期各地区气候相关程度的研究,讨论其中的演变趋势,以及全球变暖效应在其中的影响.为此选择美国过去一百年间各地观测站的气温观测值,以十年为周期分为十组,并为每组数据构造观测站相关图.每组数据都经过周期分解的预处理,以消除日照周期在其中造成的周期性影响,然后运用高斯马尔可夫随机场模型,通过对逆协方差矩阵的统计估计,得到不同观测站之间的气温序列的条件依赖关系,并以此构建相关图.因为气候模型的时间序列特殊性,并对模型的假设和拟合方式进行了改进.最后从得到的相关图中发现了一百年间相继出现的两种相反演变趋势,从而对全球变暖是其主要原因的可能性加以讨论.%Global warming is the phenomenon of average temperature increases of the Earth's near-surface air and oceans commenced from the mid-20th century.The goal of this thesis is through investigating the relationship between different regional climates from historical temperature data at different time periods to find the trends among them and to discuss the impact of the global warming on the trends.As a try of this goal,we selected the temperature observation data of past 100 years gathered by different observation stations in United States and divided them into 10 groups at the interval of 10 years each, and then built the observation stations' relevance graphs upon each of them.A seasonal decomposition procedure was used on every group as the pre-treatment to remove the seasonal effect of solar radiation on them, then the Ganssian-Markov Random Field model was employed; by statistical estimation on contra-covariance matrix the condition dependence relationship of temperature series amongst different observation stations were derived, based on this, we constructed the relevance graph.Because of the particularity of time series of climates model, again we had improved the hypothesis and fitness way of the model.At last we figured out from the derived relevance graph two opposite evolution trends occurring in succession during 100 years, and extended a discussion on the possibility of this in connection to the global warming as the main cause.
展开▼
