SMART-G Monte Carlo GPU Radiative Transfer

SMART-G (Speed-Up Monte Carlo Advanced Radiative Transfer using GPU) is an advanced radiative transfer code. Initially co-developed by HYGEOS for CNES to simulate how the light propagates in the atmosphere and ocean, SMART-G is now continuously upgraded by HYGEOS. Examples are given below.


Monte Carlo solver

Accounts for light polarization

Works in plane parallel or spherical geometry

GPU accelerated using the CUDA framework (massively parallel processing)

Accounts for reflection/transmission by a wind-roughened sea surface, including wave shadowing effects

Accounts for adjacency effects

Multispectral processing

Coded in python + pycuda

One poster presented at the IOCS conference in 2015 on few possible applications with SMART-G. Another poster on SMART-G computing time improvement by implementing the ALIS scheme. New developments in radiative transfer were also performed by HYGEOS for another code called ARTDECO. Poster presented at IRS in 2016.

Contact at Hygeos: Didier Ramon