A novel hybrid scattering order-dependent variance reduction method for Monte Carlo simulations of radiative transfer in cloudy atmosphere. (March 2017)

• Record Type: Journal Article
• Authors: Wang, Zhen; Cui, Shengcheng; Yang, Jun; Gao, Haiyang; Liu, Chao; Zhang, Zhibo
• Is Part Of: Journal of quantitative spectroscopy & radiative transfer. Volume 189(2017)
• Page Start: 283
• Page End: 302
• Publication Date: 2017-03
• Subjects: RTE radiative transfer equation -- IRTE integral radiative transfer equation -- RTM radiative transfer modeling -- FMC/BMC forward/backward Monte Carlo -- PDF probability distribution function -- CP "cloned" photon -- LEM local estimate method -- VRM variance reduction method -- PFFT phase function forward truncation -- TDIS target directional importance sampling -- OPFFT/OTDIS order-dependent PFFT/TDIS -- HSOV hybrid scattering order-dependent VRM -- TOA/BOA top/bottom of atmosphere -- VNA/VZA viewing nadir/zenith angle -- RE/RB relative error/bias -- ODRE/ODRB order-dependent RE/RB -- EF efficiency factor; Monte Carlo -- Variance reduction -- Atmosphere -- Order-dependent -- Integral radiative transfer -- Cloud; Spectrum analysis -- Periodicals; Radiation -- Periodicals; Analyse spectrale -- Périodiques; Rayonnement -- Périodiques; Radiation; Spectrum analysis; Periodicals; 543.0858
• DOI: 10.1016/j.jqsrt.2016.12.002
• Languages: English
• ISSNs: 0022-4073

Abstract

Abstract: We present a novel hybrid scattering order-dependent variance reduction method to accelerate the convergence rate in both forward and backward Monte Carlo radiative transfer simulations involving highly forward-peaked scattering phase function. This method is built upon a newly developed theoretical framework that not only unifies both forward and backward radiative transfer in scattering-order-dependent integral equation, but also generalizes the variance reduction formalism in a wide range of simulation scenarios. In previous studies, variance reduction is achieved either by using the scattering phase function forward truncation technique or the target directional importance sampling technique. Our method combines both of them. A novel feature of our method is that all the tuning parameters used for phase function truncation and importance sampling techniques at each order of scattering are automatically optimized by the scattering order-dependent numerical evaluation experiments. To make such experiments feasible, we present a new scattering order sampling algorithm by remodeling integral radiative transfer kernel for the phase function truncation method. The presented method has been implemented in our Multiple-Scaling-based Cloudy Atmospheric Radiative Transfer (MSCART) model for validation and evaluation. The main advantage of the method is that it greatly improves the trade-off between numerical efficiency and accuracy order by order. Abstract : Highlights: Forward and backward integral radiative transfer theoretical framework is developed. Hybrid scattering order-dependent variance reduction algorithm is presented. A scattering order sampling algorithm is formulated for phase function truncation. Numerical evaluation and optimization are performed order by order.