Item

Abstract

Clouds cause complex surface solar irradiance patterns of cloud shadows interspersed with cloud-enhancements regions in which irradiances exceed the clear-sky irradiance due to scattering from nearby clouds. In turn, these spatial patterns of low and high surface irradiance affect the further development of surface-driven clouds. Monte Carlo radiative transfer methods can solve the three-dimensional transfer of radiation with very high accuracy, but have long been considered too computationally expensive to be coupled to simulations of the atmosphere. Instead, atmospheric models generally resort to two-stream methods that solve the propagation of radiation only in vertical direction. Such two-stream methods are computationally relatively efficient, but unable to realistically represent the surface solar irradiance below clouds.Here, we present the first GPU-accelerated Monte Carlo radiative transfer solver that accurately solves the three-dimensional transfer of solar radiation with a technique known as ray tracing, yet is fast enough to be coupled to high-resolution cloud-resolving simulations. Using a case study of a diurnal cycle with shallow cumulus clouds developing over land, we show that accounting for the three-dimensional radiative effects of clouds results in lower cloud covers, but significantly thicker clouds compared to simulations using two-stream methods. Our Monte Carlo ray tracer enables accurate three-dimensional radiation in atmospheric models, providing new opportunities for studying the complex interactions between clouds, radiation, and the Earth’s surface.