Analysis of Seasonal Daily Pattern of CO2 Concentration in Antarctica Using Doubly Cyclic Smoothing Splines

• Activity Number: 538
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Mihoko Minami* and Ryo Kiguchi
• Companies: Keio University and Keio University
• Keywords: tensor product smooth ; doubly cyclic smoothing splines ; soft thresholding low pass fileter

Abstract

Methods: In order to flexibly estimate continuously varying daily pattern of CO2 concentration throughout a year at Syowa station in Antarctica, we propose a doubly cyclic smoothing splines method. This method is defined using the tensor product method with cyclic splines.?A doubly cyclic smoothing spline function is a univarite function of time, but the penalty is defined on a torus. For the balanced data, the basis functions can be transformed to trigonometric functions by an orthogonal basis transformation, and it is shown that this method penalizes more on higher frequency basis functions. In other words, it can be considered as a soft thresholding low pass filter in two dimension on torus. Results: We fit a linear additive model for CO2 concentration at Showa station in Antarctica with temporal trend, annual variation, wind velocity and doubly cyclic smooth term of time with daily and annual cycles. Because of a strong temporal trend and seasonal variation, it has been said that CO2 doesn't have a daily variation, but our result showed that it has a significant daily variation in summer. We also found that the effect of wind velocity on CO2 depends on season.