Theses Doctoral

Nonlinear Data Assimilation: Towards a Prediction of the Solar Cycle

Svedin, Andreas

The solar cycle is the cyclic variation of solar activity, with a span of 9-14 years. The prediction of the solar cycle is an important and unsolved problem with implications for communications, aviation and other aspects of our high-tech society. Our interest is model-based prediction, and we present a self-consistent procedure for parameter estimation and model state estimation, even when only one of several model variables can be observed.
Data assimilation is the art of comparing, combining and transferring observed data into a mathematical model or computer simulation. We use the 3DVAR methodology, based on the notion of least squares, to present an implementation of a traditional data assimilation. Using the Shadowing Filter - a recently developed method for nonlinear data assimilation - we outline a path towards model based prediction of the solar cycle. To achieve this end we solve a number of methodological challenges related to unobserved variables. We also provide a new framework for interpretation that can guide future predictions of the Sun and other astrophysical objects.


  • thumnail for Svedin_columbia_0054D_11000.pdf Svedin_columbia_0054D_11000.pdf application/pdf 7.07 MB Download File

More About This Work

Academic Units
Thesis Advisors
Spiegel, Edward A.
Ph.D., Columbia University
Published Here
March 20, 2014