Numerical Simulation of Polarized Light and Temperature in a Stratified Atmosphere with a Slowly Varying Refractive Index

Olivier Pironneau · 2024

Read Numerical Simulation of Polarized Light and Temperature in a Stratified Atmosphere with a Slowly Varying Refractive Index by Olivier Pironneau (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-11262.txt.

Introduction

Olivier Pironneau, a solo author, simulates how light and heat move through a layered atmosphere whose bending power drifts slowly, in a 2024 paper. The setting matters for greenhouse gases like carbon dioxide. The governing system is the vector radiative transfer equations, which track polarized light. When the bending power varies, the classic form breaks for some ray directions. Pironneau instead solves an integral form that avoids the bad rays.

His iterative method, Iterations on the Source, is proven monotone and convergent. It runs geometrically fast when absorption is modest, with a unique solution guaranteed. Numerical tests show what a cloud layer that bends light more than air does to temperature and scattering. The paper is a careful piece of applied mathematics aimed at climate radiation codes.

Catalog record

CREATOROlivier Pironneau
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-11262

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Numerical Simulation of Polarized Light and Temperature in a Stratified Atmosphere with a Slowly Varying Refractive Index Olivier Pironneau Abstract . This article is an attempt to elucidate the effect of a slowly varying refractive index on the temperature in a stratified atmosphere, with a particular focus on greenhouse gases such as CO 2 . It validates an iterative method for the vector radiative transfer equations (VVRTE) called Iterations on the Source. As the system proposed by Chandrasekhar and Pomraning is not well posed for all rays directions when the refractive index varies, so instead we solve an integral representation of VRTE without the problematic rays. A mathematical proof is given showing monotonicity, convergence of the iterations and existence and uniqueness. Furthermore the convergence is geometric if the absorption is not too large. Some numerical tests are performed showing the effect of a layer of cloud with a refractive index greater than air, polarisation and scattering. Introduction The investigation of polarized light in a stratified or plane parallel atmosphere provides crucial insights into the dynamics of light propagation and polarization within complex atmospheric environments; it also holds significant implications for understanding the impact of greenhouse gases on the Earth’s climate system in the presence of slightly polarizing clouds [ 3 ] . Building upon the Variable refractive index Vector Radiative Transport Equation (VVRTE), as established by the seminal works of Chandrasekhar [ 4 ] and Pomraning [ 14 ] , this article aims to elucidate numerically the effect of the refractive index on the temperature with a particular focus on the greenhouse gas (GHG) such as CO 2 . Varying concentrations of greenhouse gases are modelled by their effects on absorption and scattering coefficients in given ranges of frequencies. The study shows that the numerical method is capable of reproducing small changes due to GHG and the influence of refraction from clouds. Clouds have a refractive index close to air and varying smoothly, unlike an air/water interface for which the Fresnel laws need to be applied (see [ 16 ] , [ 6 ] and [ 10 ] for a numerical implementation). The computer graphics community has used VVRTE with varying refractive indices for realistic rendering (see [ 11 ] , [ 2 ] and the references therein). But most of their numerical implementations assume a given temperature field. Temperature variations are at the core of the present study, but the problem had also to be reformulated because the PDE (partial differential equations) of VVRTE are not well posed, due to total refraction of some rays. Hence the PDEs are converted into a set of integral equations for. the admissible rays using the method of characteristics. Iterations on the Source are then used together with Newton iterations on the temperature equations. Convergence and monotony are analysed as in [ 12 ] . The numerical implementation is more difficult because exponential integrals cannot be used, so a precision study was added. Note that this approach can be generalised in 3D (non stratified atmospheres) as in [ 7 ] , [ 13 ] . 1. Fundamental Equations In classical physics, light in a medium \Omega is an electromagnetic radiation satisfying Maxwell’s equations. The electric field {\bf E} of a monochromatic plane wave of frequency \nu propagating in direction {\bf k} , {\bf E}={\bf E_{0}}\exp({\rm i}({\bf k\cdot{\bf x}}-\nu t)) , is a solution to Maxwell equations which is suitable to describe the propagation of a ray of light in a uniform medium when \nu is very large. Such radiations are characterized by their Stokes vector {\mathbf{I}} , made of the irradiance I and 3 functions Q,U,V to define the state of polarization. The radiation source {\mathbf{F}} for an unpolarized-emitting black-body is given by the Planck function B_{\nu}(T)=\nu^{3}({\rm e}^{\frac{\nu}{T}}-1)^{-1} . The scales used in this article are defined in [ 8 ] . If {\bf E_{0}} is written in polar coordinates, {\mathbf{I}}:=\left[\begin{matrix}I\cr Q\cr U\cr V\end{matrix}\right]=\frac{1}{2}\sqrt{\frac{\epsilon_{0}}{\mu_{0}}}\left[\begin{matrix}{E_{0}}_{\theta}{E_{0}}_{\theta}^{*}+{E_{0}}_{\phi}{E_{0}}_{\phi}^{*}\cr{E_{0}}_{\theta}{E_{0}}_{\theta}^{*}-{E_{0}}_{\phi}{E_{0}}_{\phi}^{*}\cr{E_{0}}_{\theta}{E_{0}}_{\phi}^{*}-{E_{0}}_{\phi}{E_{0}}_{\theta}^{*}\cr{E_{0}}_{\phi}{E_{0}}_{\theta}^{*}-{E_{0}}_{\theta}{E_{0}}_{\phi}^{*}\end{matrix}\right]\qquad{\mathbf{F}}:=\left[\begin{matrix}\kappa_{a}B_{\nu}(T)\cr 0\cr 0\cr 0\end{matrix}\right], where \epsilon_{0} and \mu_{0} are the electric permittivity and magnetic permeability of the medium. The parameter \kappa_{a} is related to absorption and scattering (see ( 2.5 ) below) which, by the way, are quantum effects, not described by Maxwell’s equations, except as an analogy to the limit in a random media with vanishing variations [ 9 ] . Using {\mathbf{I}} rather than {\bf E}_{0} one models absorption and scattering by a PDE system known as VVRTE [ 14 ] p152, [ 16 ] , (1.1) \displaystyle\tilde{\mathbf{I}}:=\frac{{\mathbf{I}}}{n^{2}}:\quad\frac{n}{c}\partial_{t}\tilde{\mathbf{I}} \displaystyle+{\bm{\omega}}\nabla_{\bf x}\tilde{\mathbf{I}}+\frac{\nabla_{\bf x}n}{n}\cdot\nabla_{\bm{\omega}}\tilde{\mathbf{I}}+\kappa\tilde{\mathbf{I}}+{\mathbb{N}}{\mathbf{I}}-{\mathbf{I}}{\mathbb{N}} (1.2) \displaystyle=\int_{\mathbb{S}_{2}}{\mathbb{Z}}({\bf x},{\bm{\omega}}^{\prime}:{\bm{\omega}})\tilde{\mathbf{I}}\hbox{d}\omega^{\prime}+\tilde{{\mathbf{F}}}, for all {\bf x}\in\Omega,{\bm{\omega}}\in{\mathbb{S}_{2}} , where c is the speed of light, n the refractive index of the medium , {\mathbb{S}_{2}} the unit sphere, \kappa the absorption and {\mathbb{Z}} is the phase scattering matrix for rays {\bm{\omega}}^{\prime} scattered in direction {\bm{\omega}} . {\mathbb{N}} is a tensor which is related to the torsion of the characteristics of the rays [ 16 ] and which is zero for plane parallel problems. Finally, {\mathbf{F}} is the volumic source term due to the black-body radiations of air at temperature T . It is assumed that n depends weakly only, on position {\bf x}\in\Omega ; \kappa depends smoothly on {\bf x} and strongly on \nu . Because c is very large, the term \frac{1}{c}\partial_{t}{\mathbf{I}} is neglected. Thermal equilibrium is assumed because convection by wind and molecular thermal diffusion are very small so the radiation heat source in the temperature equation of the gas is zero: (1.3) \nabla_{\bf x}\cdot\int_{{\mathbb{R}}_{+}}\int_{{\mathbb{S}}_{2}}I{\bm{\omega}}\hbox{d}\omega\hbox{d}\nu=0~~\hbox{ at all points ${\bf x}\in\Omega$}. As we shall see below (Proposition 1 ), this scalar equation defines the local temperature {\bf x}\mapsto T({\bf x}) . Notation 1 . • On all variables, the tilde indicates a division by n^{2} . • To improve readability, arguments of functions are sometimes written as indices like \kappa_{\nu} and n_{z} . Following [ 11 ] , given a cartesian frame {\bf i},{\bf j},{\bf k} , the third term on the left in ( 1.1 ) is computed in polar coordinates, with {\bm{\omega}}:={\bf i}\sin\theta\cos\varphi+{\bf j}\sin\theta\sin\varphi+{\bf k}\cos\theta,\quad{\bf s}_{1}:=-{\bf i}\sin\varphi+{\bf j}\cos\varphi, \displaystyle\nabla_{\bf x}\log{n}\cdot\nabla_{\bm{\omega}}\tilde{\mathbf{I}}=\frac{1}{\sin\theta}\frac{\partial}{\partial\theta}\left\{\tilde{\mathbf{I}}(\cos\theta{\bm{\omega}}-\bm{k})\cdot\nabla_{\bf x}\log n\right\}+\frac{1}{\sin\theta}\frac{\partial}{\partial\varphi}\left\{\tilde{\mathbf{I}}~\bm{s}_{1}\cdot\nabla_{\bf x}\log n\right\}. When n does not depend on x,y but only on z , it simplifies to \nabla_{\bf x}\log{n}\cdot\nabla_{\bm{\omega}}\tilde{\mathbf{I}}=(\partial_{z}\log n)\cdot\partial_{\mu}\left\{(1-\mu^{2})\tilde{\mathbf{I}}\right\}\quad\text{ where }\mu=\cos\theta. 2. The Stratified Case For an atmosphere of thickness Z over a flat ground, the domain is \Omega={\mathbb{R}}^{2}\times\times(0,Z) , and all variables are independent of x,y . In [ 4 ] , p40-53, expressions for the phase matrix {\mathbb{Z}} are given for Rayleigh and isotropic scattering for the \varphi -averaged of I and Q , (2.1) \displaystyle\displaystyle\bar{I}:=\frac{1}{2\pi}\int_{0}^{2\pi}I\hbox{d}\varphi,\quad\bar{Q}:=\frac{1}{2\pi}\int_{0}^{2\pi}Q\hbox{d}\varphi, (2.2) \displaystyle\bar{\mathbb{Z}}_{R}=\frac{3}{2}\left[\begin{matrix}2(1-\mu^{2})(1-\mu^{\prime 2})+\mu^{2}\mu^{\prime 2}&\mu^{2}\cr\mu^{\prime 2}&1\cr\end{matrix}\right].\quad\bar{\mathbb{Z}}_{I}=\frac{1}{2}\left[\begin{matrix}1&1\cr 1&1\cr\end{matrix}\right]. The general expression of the phase matrix for Rayleigh scattering according to S. Chandrasekhar [ 4 ] is given in Appendix 9 . For a given \beta\in[0,1] , we shall consider a combination of \beta{\mathbb{Z}}_{R} (Rayleigh scattering) plus (1-\beta){\mathbb{Z}}_{I} (isotropic scatterings) [ 4 ] , [ 14 ] , [ 15 ] . As it is understood that no variable depends on \varphi , we drop the overline. The two other components of the Stokes vectors have autonomous equations, (2.3) \displaystyle\mu\partial_{z}\tilde{U}+\partial_{z}\log n(1-\mu^{2})\partial_{\mu}\tilde{U}+\kappa\tilde{U}=0, (2.4) \displaystyle\mu\partial_{z}\tilde{V}+\partial_{z}\log n(1-\mu^{2})\partial_{\mu}\tilde{U}+\kappa\tilde{V}=\frac{\mu}{2}\int_{-1}^{1}\mu^{\prime}\tilde{V}(z,\mu^{\prime})\hbox{d}\mu^{\prime}. Notation 2 . Denote the scattering coefficient a_{s}\in[0,1) , which, as \kappa (the absorption), is a function of altitude z and frequency \nu . Define (2.5) \kappa_{s}=\kappa a_{s},\qquad\kappa_{a}=\kappa-\kappa_{s}=\kappa(1-a_{s}). From ( 2.3 ),( 2.4

Read the complete text