Cloud characteristics and 3D radiative effects in EarthCARE MSI and synergy retrievals

Imaging spectrometers, such as the multispectral imager (MSI) onboard EarthCARE, are used to derive cloud properties from backscattered solar radiation. The retrievals rely on the independent column approximation and the assumption of vertically and horizontally homogeneous clouds. These 1D simplificatopns neglect the impact of cloud structure on 3D radiative transfer, leading to biases, e.g., in the derived effective radius or cloud water path of the MSI cloud produt (M-COP).

While MSI provides information on the horizontal cloud field and cloud-top structure from brightness temperatures (BTs), the active instruments of EarthCARE, the cloud profiling radar (CPR) and the atmospheric lidar (ATLID), provide vertical cloud profiles along the satellite track. In the synergy product (ACM-CAP), CPR and ATLID are combined with nadir pixels of MSI to derive best estimates of vertical atmospheric profiles, which serve as a basis for radiative transfer simulations for closure studies in the ESA EarthCARE retrieval chain. As in the single-instrument retrieval, MSI contributes to ACM-CAP under the assumption of independent columns.

In this study, cloud properties from M-COP and ACM-CAP are analyzed while accounting for cloud structure information, including cloud fraction, standard deviations, and BT gradients, which are used to identify whether a pixel is located on the sunlit or shadowy side of a cloud. By comparing sunlit and shadowy pixels, we show that 3D radiative effects introduce systematic biases in both products, with e.g., cloud water path values being higher on the sunlit side. In ACM-CAP, the magnitude of these biases depends on the relative contribution of MSI radiances to each atmospheric column and varies with cloud type and surface conditions.

Cloud properties from M-COP are compared to ACM-CAP to identify patterns of agreement and deviation, with focus on pixels for which we assume low estimated 3D bias in ACM-CAP. Radiative transfer simulations based on ACM-CAP are performed using the MYSTIC Monte Carlo solver, showing aggreement to the observations and demonstrating that the inclusion of MSI radiances in the synergy product introduces 1D/3D inconsistencies that can affect radiative closure studies.