arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1403.3306math.STstat.TH
收藏

根据目标卫星观测估算CO $_2$通量:贝叶斯方法

Estimation of CO$_2$ flux from targeted satellite observations: a Bayesian approach

Graham Cox

摘要:

我们考虑在海洋-大气界面的二氧化碳通量的估计,在垂直大气柱的混合比的加权平均。特别是,我们研究的依赖性的后验协方差的加权函数中使用的观察,动机的事实,该功能是依赖于工具,因此需要能够比较不同的权重。估计问题被认为是使用变分数据同化方法,这是承认一个等价的无限维贝叶斯公式。在我们的调查中的主要工具是一个显式的公式的先验协方差和观测算子的后验协方差。使用这个公式,我们比较加权函数集中在地球表面附近的大气层的顶部附近,在由此产生的协方差算子。我们还考虑了观测目标的问题,并询问是否有可能通过适当选择加权函数来减少规定方向的协方差。我们发现,情况并非如此-存在这样的方向,在这些方向上,人们永远无法获得信息,无论权重的选择如何。

英文摘要:

We consider the estimation of carbon dioxide flux at the ocean-atmosphere interface, given weighted averages of the mixing ratio in a vertical atmospheric column. In particular we examine the dependence of the posterior covariance on the weighting function used in taking observations, motivated by the fact that this function is instrument-dependent, hence one needs the ability to compare different weights. The estimation problem is considered using a variational data assimilation method, which is shown to admit an equivalent infinite-dimensional Bayesian formulation. The main tool in our investigation is an explicit formula for the posterior covariance in terms of the prior covariance and observation operator. Using this formula, we compare weighting functions concentrated near the surface of the earth with those concentrated near the top of the atmosphere, in terms of the resulting covariance operators. We also consider the problem of observational targeting, and ask if it is possible to reduce the covariance in a prescribed direction through an appropriate choice of weighting function. We find that this is not the case---there exist directions in which one can never gain information, regardless of the choice of weight.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑