Stokes Inversion based on Convolutional Neural Networks

Abstract

Spectropolarimetric inversions are routinely used in the field of Solar Physics for the extraction of physical information from observations. The application to two-dimensional fields of view often requires the use of supercomputers with parallelized inversion codes. Even in this case, the computing time spent on the process is still very large. Our aim is to develop a new inversion code based on the application of convolutional neural networks that can quickly provide a three-dimensional cube of thermodynamical and magnetic properties from the interpretation of two-dimensional maps of Stokes profiles. We train two different architectures of fully convolutional neural networks. To this end, we use the synthetic Stokes profiles obtained from two snapshots of three-dimensional magneto-hydrodynamic numerical simulations of different structures of the solar atmosphere. We provide an extensive analysis of the new inversion technique, showing that it infers the thermodynamical and magnetic properties with a precision comparable to that of standard inversion techniques. However, it provides several key improvements: our method is around one million times faster, it returns a three-dimensional view of the physical properties of the region of interest in geometrical height, it provides quantities that cannot be obtained otherwise (pressure and Wilson depression) and the inferred properties are decontaminated from the blurring effect of instrumental point spread functions for free. The code is provided for free on a specific repository, with options for training and evaluation.

18 pages, 14 figures, accepted for publication in Astronomy & Astrophysics

References in corpus (32)

• VFISV: Very Fast Inversion of the Stokes Vector for the Helioseismic and Magnetic Imager
• Simulation of the Formation of a Solar Active Region
• An Open Source, Massively Parallel Code for Non-LTE Synthesis and Inversion of Spectral Lines and Zeeman-induced Stokes Profiles
• Advanced Forward Modeling and Inversion of Stokes Profiles Resulting from the Joint Action of the Hanle and Zeeman Effects
• Numerical sunspot models: Robustness of photospheric velocity and magnetic field structure
• STiC -- A multi-atom non-LTE PRD inversion code for full-Stokes solar observations
• The intensity contrast of solar granulation: comparing Hinode SP results with MHD simulations
• Spatially coupled inversion of spectro-polarimetric image data I: Method and first results
• Probing quiet Sun magnetism using MURaM simulations and Hinode/SP results: support for a local dynamo
• Structure of sunspot penumbral filaments: a remarkable uniformity of properties
• Why one-dimensional models fail in the diagnosis of average spectra from inhomogeneous stellar atmospheres
• DeepVel: deep learning for the estimation of horizontal velocities at the solar surface
• Evidence for Quasi-isotropic Magnetic Fields from Hinode Quiet Sun Observations
• On the validity of the 630 nm Fe I nm lines for the magnetometry of the internetwork quiet Sun
• Returning magnetic flux in sunspot penumbrae
• Enhancing SDO/HMI images using deep learning
• Three-dimensional structure of a sunspot light bridge
• Real-time multiframe blind deconvolution of solar images
• A geometrical height scale for sunspot penumbrae
• Segmentation of coronal holes in solar disk images with a convolutional neural network
• Sparse inversion of Stokes profiles. I. Two-dimensional Milne-Eddington inversions
• Strategies for Spectral Profile Inversion using Artificial Neural Networks
• Spatial deconvolution of spectropolarimetric data: an application to quiet Sun magnetic elements
• Spectropolarimetric analysis of an active region filament. I. Magnetic and dynamical properties from single component inversions
• A new MHD-assisted Stokes inversion technique
• The intrinsic dimensionality of spectro-polarimetric data
• Zeeman-Tomography of the Solar Photosphere -- 3-Dimensional Surface Structures Retrieved from Hinode Observations
• Measuring the Wilson depression of sunspots using the divergence-free condition of the magnetic field vector
• RADYNVERSION: Learning to Invert a Solar Flare Atmosphere with Invertible Neural Networks
• Diagnostic potential of the Ca II 8542A line for solar filaments
• Spectropolarimetric analysis of an active region filament. II. Evidence of the limitations of a single component model
• A GPU-Computing Approach to Solar Stokes Profile Inversion

Cited by in corpus (31)

• Waves in the lower solar atmosphere: the dawn of next-generation solar telescopes
• Solar image denoising with convolutional neural networks
• Machine learning initialization to accelerate Stokes profile inversions
• Bayesian Stokes inversion with Normalizing flows
• Inferring Vector Magnetic Fields from Stokes Profiles of GST/NIRIS Using a Convolutional Neural Network
• Spatio-temporal analysis of chromospheric heating in a plage region
• Measuring the magnetic origins of solar flares, CMEs and Space Weather
• Mimicking spectropolarimetric inversions using convolutional neural networks
• Physical properties of bright Ca II K fibrils in the solar chromosphere
• Combining magneto-hydrostatic constraints with Stokes profile inversions. II. Application to Hinode/SP observations
• SynthIA: A Synthetic Inversion Approximation for the Stokes Vector Fusing SDO and Hinode into a Virtual Observatory
• Novel framework for the three-dimensional NLTE inverse problem
• Accurately constraining velocity information from spectral imaging observations using machine learning techniques
• Classification of High-resolution Solar Hα Spectra using t-distributed Stochastic Neighbor Embedding
• Exploring spectropolarimetric inversions using neural fields. Solar chromospheric magnetic field under the weak-field approximation
• Inference of electric currents in the solar photosphere
• A Non-Linear Magnetic Field Calibration Method for Filter-Based Magnetographs by Multilayer Perceptron
• SunnyNet: A neural network approach to 3D non-LTE radiative transfer
• Spectropolarimetric Inversion in Four Dimensions with Deep Learning (SPIn4D): I. Overview, Magnetohydrodynamic Modeling, and Stokes Profile Synthesis
• Inferring Line-of-Sight Velocities and Doppler Widths from Stokes Profiles of GST/NIRIS Using Stacked Deep Neural Networks
• Connecting the Wilson depression to the magnetic field of sunspots
• A nonlinear solar magnetic field calibration method for the filter-based magnetograph by the residual network
• A study of the capabilities for inferring atmospheric information from high-spatial-resolution simulations
• Mapping the Sun's upper photosphere with artificial neural networks
• Convolutional Neural Networks and Stokes Response Functions
• On the effect of oscillatory phenomena on Stokes inversion results
• The MODEST catalog of depth-dependent spatially coupled inversions of sunspots observed by Hinode/SOT-SP
• Spectral resolution effects on the information content in solar spectra
• Neural translation for Stokes inversion and synthesis
• Solar multi-object multi-frame blind deconvolution with a spatially variant convolution neural emulator
• Compression Method for Solar Polarization Spectra Collected from Hinode SOT/SP Observations