Assessing tools for 3-dimensional radiative transfer

Many scientists who want to solve three-dimensional radiative transfer problems are not themselves expert in the underlying theory, so they use whatever tools they can find to solve the problem. The best-documented and most widely-available of these tools is an implementation of the Spherical Harmonics Discrete Ordinates (SHDOM) method that uses adaptive mesh refinement to solve the complete radiative transfer problem throughout the problem domain. Radiative transfer practitioners know that Monte Carlo (MC) methods are substantially more efficient for some class of problems. Several MC implementations are also available, including the recently-released I3RC (Intercomparison of Three-dimensional Radiation Codes) Community Monte Carlo model, which is explicitly intended for use by the broader scientific community. How are users to know which method makes the most sense for their problem?

Here we compare the computational costs of computing three-dimensional radiative transfer in plane-parallel atmospheres using SHDOM and the I3RC Community Monte Carlo model. We solve several realistic problems with applications to remote sensing, energy budgets, and local heating rates. We examine the memory and CPU time required by each algorithm as a function of the desired accuracy, and note where the problems sizes and/or accuracy requirements at which each algorithm becomes untenable. The results demonstrate the trade-offs associated with each algorithm and provide a rough guide as to when the choice is clear.