Abstract
Magnetic fields are responsible for a multitude of solar phenomena, including potentially destructive events such as solar flares and coronal mass ejections, with the number of such events rising as we approach the peak of the 11-year solar cycle in approximately 2025. High-precision spectropolarimetric observations are necessary to understand the variability of the Sun. The field of quantitative inference of magnetic field vectors and related solar atmospheric parameters from such observations has been investigated for a long time. In recent years, very sophisticated codes for spectropolarimetric observations have been developed. Over the past two decades, neural networks have been shown to be a fast and accurate alternative to classic inversion methods. However, most of these codes can be used to obtain point estimates of the parameters, so ambiguities, degeneracies, and uncertainties of each parameter remain uncovered. In this paper, we provide end-to-end inversion codes based on the simple Milne-Eddington model of the stellar atmosphere and deep neural networks to both parameter estimation and their uncertainty intervals. The proposed framework is designed in such a way that it can be expanded and adapted to other atmospheric models or combinations of them. Additional information can also be incorporated directly into the model. It is demonstrated that the proposed architecture provides high accuracy results, including a reliable uncertainty estimation, even in the multidimensional case. The models are tested using simulations and real data samples.
Similar content being viewed by others
Data Availability
In the current study, we used a collection of the Level 1 calibrated Stokes spectra (comprised by images stored in FITS format) and a collection of the Level 2 data sets (obtained from the MERLIN spectral line inversion of the Level 1 calibrated spectra) produced by the Spectropolarimeter (SP) on board the Hinode, since its launch in 2006 (collected in the Community Spectropolarimetric Analysis Center (CSAC) at HAO/NCAR). Hinode is a Japanese mission, developed and launched by ISAS/JAXA, with NAOJ as a domestic partner and NASA and STFC (UK) as international partners. It is operated by these agencies in cooperation with ESA and NSC (Norway). The Hinode has an open data policy, allowing anyone to access the data and data products. Level 1 and 2 data are available by following the data link.
References
Baso, C.J.D., Ramos, A.A., de la Cruz Rodríguez, J.: 2022, Bayesian Stokes inversion with normalizing flows. Astron. Astrophys. 659, A165. DOI.
Blundell, C., Cornebise, J., Kavukcuoglu, K., Wierstra, D.: 2015, Weight uncertainty in neural networks. J. Mach. Learn. Res.. DOI.
Borrero, J.M., Tomczyk, S., Kubo, M., Socas-Navarro, H., Schou, J., et al.: 2011, VFISV: very fast inversion of the Stokes vector for the helioseismic and magnetic imager. Solar Phys. 273(1), 267. DOI.
Carroll, T., Kopf, M., Strassmeier, K.: 2008, A fast method for Stokes profile synthesis. Astron. Astrophys. 488(2), 781. DOI.
Carroll, T.A., Staude, J.: 2001, The inversion of Stokes profiles with artificial neural networks. Astron. Astrophys. 378, 316. DOI.
Dalda, A.S., de la Cruz Rodríguez, J., Pontieu, B.D., Gošić, M.: 2019, Recovering thermodynamics from spectral profiles observed by IRIS: a machine and deep learning approach. Astrophys. J. Lett. 875(2), L18. DOI.
del Toro Iniesta Carlos, J., Ruiz Cobo, B.: 2016, Inversion of the radiative transfer equation for polarized light. Living Rev. Solar Phys. 13(1), 4. DOI.
Gafeira, R., Suárez, D.O., Milić, I., Noda, C.Q., Cobo, B.R., Uitenbroek, H.: 2021, Machine learning initialization to accelerate Stokes profile inversions. Astron. Astrophys. 651, A31. DOI.
Gawlikowski, J., Cedrique Rovile, N.T., Ali, M., Lee, J., Humt, M., Feng, J., et al.: 2021, A survey of uncertainty in deep neural networks. Preprint. arXiv. DOI.
Ghahramani, Z.: 2015, Probabilistic machine learning and artificial intelligence. Nature 521(7553), 452. DOI.
Guo, J., Bai, X., Liu, H., Yang, X., Deng, Y., Lin, J., et al.: 2021, A nonlinear solar magnetic field calibration method for the filter-based magnetograph by the residual network. Astron. Astrophys. 646, A41. DOI.
Higgins, R.E.L., Fouhey, D.F., Zhang, D., Antiochos, S.K., Barnes, G., Todd, H.J., et al.: 2021, Fast and accurate emulation of the SDO/HMI Stokes inversion with uncertainty quantification. Astrophys. J. 911(2), 130. DOI.
Higgins, R.E.L., Fouhey, D.F., Antiochos, S.K., Barnes, G., Cheung, M.C.M., Todd, H.J., et al.: 2022, SynthIA: a synthetic inversion approximation for the Stokes vector fusing SDO and hinode into a virtual observatory. Astrophys. J. Suppl. 259(1), 24. DOI.
Jiang, H., Li, Q., Xu, Y., Hsu, W., Ahn, K., Cao, W., Wang, J.T.L., Wang, H.: 2022, Inferring line-of-sight velocities and Doppler widths from Stokes profiles of gst/niris using stacked deep neural networks. Astrophys. J. 939(2), 66. DOI.
Knyazeva, I., Plotnikov, A., Medvedeva, T., Makarenko, N.: 2022, Multi-output deep learning framework for solar atmospheric parameters inferring from Stokes profiles. In: Advances in Neural Computation, Machine Learning, and Cognitive Research V 1008, Springer, Cham, 299. DOI.
Kostenetskiy, P.S., Chulkevich, R.A., Kozyrev, V.I.: 2021, HPC resources of the higher school of economics. J. Phys. Conf. Ser. 1740(1), 012050. DOI.
Krzywinski, M., Altman, N.: 2013, Power and sample size. Nat. Methods 10(12), 1139. DOI.
Kuckein, C., Balthasar, H., Noda, C.Q., Diercke, A., Arjona, J.C.T., Cobo, B.R., et al.: 2021, Multiple Stokes I inversions for inferring magnetic fields in the spectral range around Cr I 5782 Å. Astron. Astrophys. 653, A165. DOI.
Kuleshov, V., Fenner, N., Ermon, S.: 2018, Accurate uncertainties for deep learning using calibrated regression. DOI.
Lakshminarayanan, B., Pritzel, A., Blundell, C.: 2017, Simple and scalable predictive uncertainty estimation using deep ensembles. In: Advances in Neural Information Processing Systems, MIT Press, Cambridge, 6405. DOI.
Landi Degl’Innocenti, E., Landolfi, M.: 2004, Polarization in Spectral Lines, Springer, Dordrecht. DOI.
Leka, K.D., Wagner, E.L., Griñón-Marín, A.B., Bommier, V., Higgins, R.E.L.: 2022, On identifying and mitigating bias in inferred measurements for solar vector magnetic field data. Solar Phys. 297, 121. DOI.
Li, H., Xu, Z., Qu, Z., Sun, L.: 2019, MCMC inversion of Stokes profiles. Astrophys. J. 875(2), 127. DOI.
Li, H., del Pino Alemán, T., Bueno, J.T., Casini, R.: 2022, TIC: a Stokes inversion code for scattering polarization with partial frequency redistribution and arbitrary magnetic fields. Astrophys. J. 933(2), 145. DOI.
Lites, B.W., Ichimoto, K.: 2013, The SP_PREP data preparation package for the hinode spectro-polarimeter. Solar Phys. 283(2), 601. DOI.
Lites, B., Casini, R., Garcia, J., Socas-Navarro, H.: 2006, A suite of community tools for spectro-polarimetric analysis. MemSAIt 78, 148.
Liu, H., Xu, Y., Wang, J., Jing, J., Liu, C., Wang, J.T.L., Wang, H.: 2020, Inferring vector magnetic fields from Stokes profiles of GST/NIRIS using a convolutional neural network. Astrophys. J. 894(1), 70. DOI.
Malinin, A., Gales, M.: 2018, Predictive uncertainty estimation via prior networks. In: Advances in Neural Information Processing Systems, Curran Associates, Red Hook. DOI.
Milić, I., Gafeira, R.: 2020, Mimicking spectropolarimetric inversions using convolutional neural networks. Astron. Astrophys. 644, A129. DOI.
Okamoto, T., Tsuneta, S., Lites, B., Kubo, M., Yokoyama, T., Berger, T.E., et al.: 2009, Prominence formation associated with an emerging helical flux rope. Astrophys. J. 697, 913. DOI.
Pawitan, Y.: 2001 In All Likelihood: Statistical Modelling and Inference Using Likelihood, OUP, Oxford.
Podladchikova, T., Jain, S., Veronig, A., Sutyrina, O., Dumbovic, M., Clette, F., et al.: 2022, Maximal growth rate of the ascending phase of a sunspot cycle for predicting its amplitude. Astron. Astrophys. 663, A88. DOI.
Quinonero-Candela, J., Rasmussen, C.E., Sinz, F., Bousquet, O., Schölkopf, B.: 2005, Evaluating Predictive Uncertainty Challenge, Machine Learning Challenges Workshop 1, Springer, Berlin. DOI.
Ramos, A.A., Baso, C.D.: 2019, Stokes inversion based on convolutional neural networks. Astron. Astrophys. 626, A102. DOI.
Ramos, A.A., González, M.J.M., Rubiño-Martín, J.A.: 2007, Bayesian inversion of Stokes profiles. Astron. Astrophys. 476(2), 959. DOI.
Ramos, A.A., de la Cruz Rodríguez, J., González, M.J.M., Yabar, A.P.: 2016, Inversion of Stokes profiles with systematic effects. Astron. Astrophys. 590, A87. DOI.
Ramos, A.A., de la Cruz Rodríguez, J., González, M.J.M., Socas-Navarro, H.: 2017, Inference of the chromospheric magnetic field orientation in the ca ii 8542 a line fibrils. Astron. Astrophys. 599, A133. DOI.
Rudenko, G.V., Anfinogentov, S.A.: 2013, Very fast and accurate azimuth disambiguation of vector magnetograms. Solar Phys. 289(5), 1499. DOI.
Shorten, C., Khoshgoftaar, T.M.: 2019, A survey on image data augmentation for deep learning. J. Big Data 6(1), 60. DOI.
Shrestha Durga, L., Solomatine Dimitri, P.: 2006, Machine learning approaches for estimation of prediction interval for the model output. Neural Netw. 19(2), 225. DOI.
Socas-Navarro, H.: 2005, Strategies for spectral profile inversion using artificial neural networks. Astrophys. J. 621(1), 545. DOI.
Unno, W.: 1956, Line formation of a normal Zeeman triplet. Publ. Astron. Soc. Japan 8, 108.
Viticchié, B., Almeida, J.S.: 2011, Asymmetries of the Stokes V profiles observed by Hinode SOT/SP in the quiet sun. Astron. Astrophys. 530, A14. DOI.
Xu, W., Chen, W., Liang, Y.: 2018, Feasibility study on the least square method for fitting non-Gaussian noise data. Physica A 492, 1917. DOI.
Acknowledgments
Denis Derkach, Lukia Mistryukova, Aleksandr Khizhik, and Mikhail Hushchyn are grateful to the HSE basic research program. This research was supported in part through computational resources of HPC facilities at HSE University (Kostenetskiy, Chulkevich, and Kozyrev 2021).
Author information
Authors and Affiliations
Contributions
D.D. conceived and led the research; L.M. and A.K. performed calculations and processed the results; A.P., I.K., M.H. and D.D. analyzed and discussed the results; L.M., I.K. and D.D. wrote the main manuscript text; A.P. and I.K. processed data samples. All authors reviewed the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Mistryukova, L., Plotnikov, A., Khizhik, A. et al. Stokes Inversion Techniques with Neural Networks: Analysis of Uncertainty in Parameter Estimation. Sol Phys 298, 98 (2023). https://doi.org/10.1007/s11207-023-02189-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11207-023-02189-4