Improving Solar Dynamo Modeling

Wednesday, February 11, 2015

Approximating a model’s “true” state by applying Kalman filtering to solar cycles, a technique first developed in the 1950s and 1960s to help Apollo 11 settle safely on the Moon.

Recent work by HAO scientist Mausumi Dikpati1, and CISL scientist Jeffrey L. Anderson, in collaboration with NORDITA scientist Dhrubaditya Mitra, used data assimilation to examine several questions at the heart of solar cycle prediction, including the number and type of observations needed to constrain the problem, and the implications of using incorrect model ingredients.

Kalman filtering data assimilation modeling image
By applying Kalman filtering data assimilation to synthetic solar flow data, the authors found that they could approximate the model’s “true” state. (Geophysical Research Letters, doi:10.1002/2014GL061077, 2014)

Being able to predict the impact of a single storm is important, but it is also important to predict when we are in for a particularly stormy season. Not only is it difficult to predict exactly when solar activity will reach a maximum, but it is also a challenge to predict just how active a maximum is likely to be. A key problem is the lack of consensus about the Sun's meridional circulation pattern.

By implementing a sophisticated Ensemble Kalman filter data assimilation technique, using an NCAR-based DART (Data Assimilation Research Testbed) tool, Dikpati et al. assimilated surface observations to approximate the “true” state of the meridional circulation in the convection zone below the Sun's surface. Kalman filtering was first developed in the late 1950s and 1960s, Kalman filtering is a signal-processing approach used to create a rolling best estimate of a system’s properties when the needed observations are drowned by noise. The technique was a significant boon to early space missions: Computers equipped with Kalman filters helped the Apollo 11 module settle safely on the Moon. For details of implementation of the technique to solar cycle models, click AGU's Research Spotlight, click HAO 2014 Annual Report on Data Assimilation.