Mechanisms of surface solar irradiance variability under broken clouds

Surface solar irradiance variability is present under all broken clouds, but the patterns, magnitude of variability, and mechanisms behind it vary greatly with cloud type. Most radiative transfer models do not resolve the observed variability, primarily due to limiting radiative transport to two streams (up and down) to save computation time. From observations, we selected a diverse set of surface solar irradiance patterns under various cloud types and modelled these cloud types in combination with a Monte Carlo ray tracer for accurate 3D radiative transfer. Stratus, altocumulus, and cumulus growing into cumulonimbus are among the studied cloud types. The goal of these experiments is to understand through which mechanisms various cloud types generate observed patterns of irradiance variability. 

The results show that we can capture the essence in four mechanisms. We find that for optically thin (optical thickness  <  6) clouds, scattering in the forward direction dominates. In cloud fields with enough optically thin area, such as altocumulus, "forward escape" alone can drive areas of irradiance enhancement of over 50 % of clear-sky irradiance.  For flat, optically thick clouds (optical thickness > 6), irradiance is instead scattered diffusely downward ("downward escape"), and (extreme) enhancements are thus found directly below the cloud rather than in the direction along the solar angle. For vertically structured clouds, "side escape" dominates domain-averaged diffuse irradiance enhancement until anvil clouds form and start shading the updraft.  Lastly, under optically thick cloud cover, surface albedo enhances downward radiative fluxes due to multiple scattering between surface and cloud. This both brightens shadows and contributes 10 to 60 % of the total irradiance enhancement in sunlit areas for respectively low (0.2) to high (0.8) albedo. 

With these four mechanisms, we provide a framework for understanding the vast diversity and complexity found in surface solar irradiance and cloudiness. Such a framework can guide the development of parameterizations that capture 3D solar irradiance effect at a fraction of the computational cost of 3D radiative transfer models, for example.