A three-dimensional parallel radiative transfer model for atmospheric heating rates for use in cloud resolving models-The TenStream solver

- We present a fast 3D radiative transfer solver for atmospheric models.
- New solver accurately approximates 3D RT effects.
- Consistently reduces the error compared to 1D solvers.
- Performance is several orders of magnitude better than for Monte Carlo solvers.
- Parallelized solver is suitable for use in LES and high resolution NWP models.

This paper presents a new method to compute three-dimensional heating rates in atmospheric models, in particular numerical weather prediction models and large eddy simulations. The radiative transfer in such models is usually calculated for each vertical column independent of its neighbouring columns. Earlier studies showed that the neglect of horizontal energy transport introduces significant errors at model grid spacings below 1 km. To date, there is no method to calculate 3D heating rates which is fast enough to systematically study the effect of radiation on cloud evolution. Here, we present a new algorithm which provides a fast yet accurate approximation for realistic three-dimensional heating rates. The method extends the well-known one-dimensional two-stream theory to 10 streams in three dimensions. Special emphasis is laid on scalable parallelism and speed.It is found that the new solver significantly reduces the root mean square error for atmospheric heating and surface heating rates when compared to traditionally employed one-dimensional solvers. The TenStream solver reduces the relative root mean square error of heating rates by a factor of five when compared to the independent column approximation. In the case of a strato-cumulus cloud field and the solar zenith angle being 60°, the error was reduced from 178% to 31% and for a deep-convective cumulus cloud from 138% to 28%. The model described here will open the way to answer the question, if and how much three-dimensional radiative transfer effects indeed affect cloud development and precipitation.