Sparse Bayesian Inference and the Temperature Structure of the Solar Corona

TL;DR

This paper applies sparse Bayesian inference to infer the temperature distribution of the solar corona from spectroscopic data, providing a regularized, less oscillatory solution compared to traditional methods.

Contribution

It introduces a Bayesian framework with sparsity priors for more stable and physically plausible temperature inferences in solar atmospheric studies.

Findings

Effective in reducing oscillatory solutions

Can handle a library of assumed temperature distributions

Avoids ad hoc assumptions in inversion process

Abstract

Measuring the temperature structure of the solar atmosphere is critical to understanding how it is heated to high temperatures. Unfortunately, the temperature of the upper atmosphere cannot be observed directly, but must be inferred from spectrally resolved observations of individual emission lines that span a wide range of temperatures. Such observations are "inverted" to determine the distribution of plasma temperatures along the line of sight. This inversion is ill-posed and, in the absence of regularization, tends to produce wildly oscillatory solutions. We introduce the application of sparse Bayesian inference to the problem of inferring the temperature structure of the solar corona. Within a Bayesian framework a preference for solutions that utilize a minimum number of basis functions can be encoded into the prior and many ad hoc assumptions can be avoided. We demonstrate the efficacy of the Bayesian approach by considering a test library of 40 assumed temperature distributions.