The state of Earth's atmosphere and our observation of it are both determined by electromagnetic radiation that transports energy and information. So, whether we want to retrieve atmospheric variables from remote sensing, predict the climatic development of Earth's temperature or understand how temperature gradients drive weather dynamics, we need to calculate atmospheric radiative transport. One of the major challenges is the strong interaction between radiation and clouds, as clouds develop in complex 3D shapes and change rapidly in both space and time.

Figure 1: 3D radiative transport through an inhomogeneous cloud field in nature Modelling this complexity exactly is numerically expensive and needs 3D observations that have, until very recently, not been available. Hence operational radiation schemes in both observation retrieval and global weather and climate models have mostly - unrealistically - assumed that clouds are horizontally homogeneous and therefore a 1-dimensional vertical calculation is sufficient. With model resolutions rapidly increasing, this project adresses the need to better understand the 3D aspects of radiation and cloud structure, and improve cloud-radiation parameterisations in global models.

High-resolution experiments using large-eddy simulations and full 3D Monte Carlo radiation calculations allow us to pinpoint the cloud geometry parameters that determine 3D radiative transport: cloud edge length and inhomogeneity, in-cloud radiation distribution and clustering of separate clouds. For real clouds, we derive cloud structure parameters from ground-based scanning cloud radar data at the Jülich Observatory for Cloud Evolution (JOYCE) as well as space-bourne A-train satellite observations.


To incorporate the 3D effects we observe in a computationally efficient way into global models, we have developed a new radiation scheme, the Speedy Algorithm for Radiative Transfer through Cloud Sides (SPARTACUS), and implemented it in a version of ECMWF's radiation model. Evaluation with 3D Monte Carlo calculations shows that SPARTACUS captures 3D effects in both longwave and shortwave well, at a fraction of the cost of full 3D models, making it possible to analyse 3D cloud effects globally.

We find significant global 3D effects on cloud-radiation interaction in both shortwave and longwave. The 3D effect on net CRF ranges over the globe from -44 Wm-2 to +26 Wm-2 , where the sign of the change depends on solar zenith angle and cloud type. Through changes in heating rates 3D effects also feed back on cloud development.