SPATIAL REGRESSION METHODS IN DENDROCLIMATOLOGY - A REVIEW AND COMPARISON OF 2 TECHNIQUES
SPATIAL REGRESSION METHODS IN DENDROCLIMATOLOGY - A REVIEW AND COMPARISON OF 2 TECHNIQUES
复制标题
DOI：
10.1002/joc.3370140404
复制
发表时间：
1994-05-01
期刊：
影响因子：
--
通讯作者：
JONES, PD
中科院分区：
文献类型：
--
作者：
COOK, ER;BRIFFA, KR;JONES, PD
我们回顾和比较两种替代的空间回归方法在树木年轮气候学重建气候。这些方法是正交空间回归（OSR）和典型回归（CR）。OSR和CR方法在最小二乘理论中具有共同的基础，并且当所有p个候选树轮气候预测因子被强制进入模型时，它们收敛到相同的解。然而，当仅使用子集p' < p预测器时，OSR和CR的性能可能不同。理论无法预测当应用最佳子集选择时，任何一种方法可能如何执行，特别是关于重建精度。因此，OSR和CR的经验比较，使用三个树木年轮和气候网络，从西欧和北美东部已被用于以前的树木气候研究。这些比较依赖于一套回归模型验证统计数据，以验证最佳子集模型产生的气候重建的准确性。结果表明，OSR和CR之间的真实的差异很小，每一个表现同样好或坏，这取决于树木年轮中可恢复的气候信息量。典型回归在高信噪比的情况下可能表现得稍好;相反，OSR在信噪比较低时可能表现得稍好。然而，这些明显的差异都不足以选择一种方法优先于另一种方法，需要进行更多的比较才能确定这些迹象是否普遍有效。
We review and compare two alternative spatial regression methods used in dendroclimatology to reconstruct climate from tree rings. These methods are orthogonal spatial regression (OSR) and canonical regression (CR). Both the OSR and CR methods have a common foundation in least-squares theory and converge to the same solution when all p candidate tree-ring predictors of climate are forced into the model. However, the performance of OSR and CR may differ when only subsets p' < p predictors are used. Theory cannot predict how either method is likely to perform when best-subset selection is applied, especially with regards to reconstruction accuracy. Consequently, empirical comparisons of OSR and CR are made using three tree-ring and climate networks from western Europe and eastern North America that have been used in previous dendroclimatic studies. These comparisons rely on a suite of regression model verification statistics to validate the accuracy of the climatic reconstructions produced by the best-subset models. The results indicate little real difference between OSR and CR, with each performing equally good or bad depending on the amount of recoverable climatic information in the tree rings. Canonical regression may perform slightly better in high signal-to-noise cases; conversely, OSR may perform slightly better when the signal-to-noise ratio is low. None of these apparent differences are large enough to select one method in preference to the other, however, and many more comparisons would be needed to determine if such indications are generally valid.