A fast open source code for determining the intensity and angular distribution of radiation transmitted through homogenous cloud cover

Predicting the transmission of solar radiation through clouds is critical to forecasting the performance of solar power systems. The interaction of solar radiation with cloud particles is described by Mie scattering and the subsequent radiation transport equations must be solved numerically. A standard implementation is typically done using a discrete ordinates or Monte Carlo approach, both of which can be computationally expensive. An alternate approach, called the doubling-method, approximates a cloud as being comprised of optically thin layers for which transmission and reflection functions can be formulated. A thick cloud is then modeled by stacking the layers in such a way that the reflection and transmission from one layer act as inputs to the layers above and below it. The approach allows for rapid computation of the direct and diffuse components of surface solar radiation as a function of solar zenith angle, wavelength, and cloud optical thickness. Here we present a simple, platform independent, fully benchmarked, algorithm for implementing the doubling-method. The code runs orders of magnitude faster than does DISORT for an equivalent angular resolution and is shown to produce equivalent results.