A Method for Estimating Potential Seasonal Predictability: Analysis of Covariance
A Method for Estimating Potential Seasonal Predictability: Analysis of Covariance
复制标题
DOI：
10.1175/jcli-d-11-00342.1
复制
发表时间：
2012-08
影响因子：
4.9
通讯作者：
Xia Feng;T. DelSole;P. Houser
中科院分区：
文献类型：
--
作者：
Xia Feng;T. DelSole;P. Houser
本文提出了一种新的方法来评估潜在的可预测性的季节性手段，使用一个单一的实现每天的时间序列。潜在的可预报性被定义为季节性的变化，超过了由于天气随机过程的变化。所提出的方法是基于协方差分析和帐户的自相关性在每日的时间序列和统计参数的不确定性。该方法适用于重新分析每日地表气温，并检测海洋和赤道陆地地区的重要潜在可预测性。潜在的可预测性较弱，并随季节变化显着，在热带以外的土地地区，与潜在的可预测的方差很少超过60%的分数。所提出的方法还产生了一个估计的潜在的可预测的季节性手段，它可以用来调查潜在的可预测性和可能的边界强迫之间的关系。结果与以前的研究基本一致，但更详细的研究将在未来的论文。
This paper proposes a new method for assessing potential predictability of seasonal means using a single realization of daily time series. Potential predictability is defined as variability in seasonal means that exceeds the variability due to weather stochastic processes. The proposed method is based on analysis of covariance and accounts for autocorrelation in daily time series and uncertainties in statistical parameters. The method is applied to reanalyzed daily surface air temperature and detects significant potential predictability over the oceans and equatorial land areas. Potential predictability is weaker and varies significantly with season over extratropical land areas, with the fraction of potentially predictable variance rarely exceeding 60%. The proposed method also produces an estimate of the potentially predictable component of seasonal means, which can be used to investigate the relation between potential predictability and possible boundary forcings. The results are generally consistent with previous studies, although a more detailed study will be made in a future paper.