应用方差分析、经验正交函数、合成分析，研究了天目山柳杉(Cryptomeria fortunei)树轮中 随方位的分布特征。应用相关分析探讨了这种分布特征的年际差异与环境因素的关系。结果表明：树轮 值有显著方位差异和年际变化，不同方位的差异可以用4个基本角分布型表示，其中一型表示各方位 C比多年平均值偏大，其中北（N）、西（W）、西南（SW）、南（S）较平均值偏大约0.5，而其余4个方位偏大不到0.2；二型各方位也较多年平均值偏大，但各方位差异不大，除N方位较小外，其余均在0.2～0.4之间。三型为各方位 较平均状况偏小，其中西北（NW）、东北（NE）方位约偏小0.4，其余方位偏小0.2～0.3。四型表现为NE、东（E）、东南（SE）与平均状况一致，西北（NW）方位较平均状况偏大约0.1，N方位偏小0.2，其余3个方位则偏小约0.3。而其年际变化既可用这4个角分布型的更替表示，也可以用第一、二主成分表示。 的第一主成分与前一年8～12月降水量负相关，与当年1～3月降水正相关。第二主成分与前一年8月至当年1月的6个月平均气温正相关。不同的角分布型与不同的气候状态相联系。一型出现的有利气候条件是：R8～12少、R1～3多、T3～7低并且T8～1高；二型出现的有利气候条件是：R8～12少、R1～3多、T3～7低并且T8～1低；三型出现的有利气候条件是：R8～12多、R1～3少、T3～7高并且T8～1高；四型出现的有利气候条件是：R8～12多、R1～3少、T3～7高并且T8～1低。可见角分布型的差异可能包含着环境变化的信息，因此同时利用树轮整体平均 值的序列和角分布型序列来重建历史气候，可以得到更多的历史气候变化信息。
The characteristics of azimuth distribution of in Cryptomeria fortunei tree rings in Tianmu Mountain regions was studied by means of statistic methods including variance analysis，composition analysis and empirical orthogonal function (EOF) analysis. The effects of environmental factors on the interannual alternation of azimuth distribution patterns were also investigated by correlation analysis. The results showed that in Cryptomeria tree rings aried along with both azimuth and time significantly. The variation along with azimuth can be represented by four basic azimuth distribution types (denoted by Types 1,2,3 and 4). Type 1 : the anomalies in all azimuths are positive, and they are >0.5 in north (N), west (W), southwest (SW) and south (S) and between 0.2-0.4 in the other four directions. Type 2 : also with positive anomalies in all azimuths as in Type 1, but the anomalies are in the interval 0.2-0.4 in most directions except for N (<0.2). Type 3 : the anomalies are about -0.4 in the NW and NE, and in the range of -0.3 to -0.2 in all other directions. Type 4 : the anomalies in NE, E and SE directions are near 0.0 while at 0.1, in the NW, -0.2 in N, and -0.3 in the other 3 directions. The variation with time can be presented by either the alternation of the four basic types or by the first two principle components (PC1 and PC2) of the eight-azimuth time-series. Furthermore, the PC1 significantly positively correlated with the total precipitation for the period from January to March and negatively correlated with the total rainfall for the period from August to December of the last year. There was a positive correlation between average temperature for the period from August to January along the PC2 axis. The azimuth distribution type varies under different climatic conditions. The favorable environmental factors for types 1 and 2 include: 1) shortage of precipitation from August to December of the preceding year; 2) richness of rainfall from January to March, and 3) lower temperature from March to July and higher (lower) temperature from last August to present January. The opposite climatic conditions were favorable to types 4 and 3. This phenomenon suggests that the azimuth distribution types may contain environmental information, which could be used to established more detailed historical climate data by both averaged in whole tree ring and its azimuth distribution type time series.