Advancements in remote sensing of the thermodynamic cloud phase using Meteosat satellites

Abstract

Clouds can consist entirely of liquid droplets, ice crystals or a mixture of both, called mixed-phase. Knowledge of the thermodynamic cloud phase is crucial for understanding the Earth’s radiation budget, cloud and atmospheric processes, and the water cycle. However, the microphysical processes governing cloud phase and phase transitions are not well understood, leading to large uncertainties in climate predictions. Especially mixed phase clouds still pose a challenge. To date, the most reliable methods for cloud phase determination from satellites are synergistic lidar-radar techniques, such as the DARDAR (liDAR-raDAR) product. But active remote sensing is limited by its narrow field of view and low temporal resolution. These missing pieces can be provided by geostationary passive sensors. However, passive remote sensing of cloud phase is challenging and remote sensing of the more complex mixed-phase in particular is rarely done. This study addresses these challenges and provides a comprehensive analysis of the phase detection capabilities of the SEVIRI instrument aboard the geostationary Meteosat Second Generation satellite. First, an analysis of the geographic and temporal distribution of cloud phases on the SEVIRI disc using the reliable DARDAR data as "ground truth" shows that all cloud phases are relevant for SEVIRI, including the mixed-phase. Second, the information content of infrared-window brightness temperature differences (BTDs) of SEVIRI is investigated. Sensitivities of the BTDs to all radiatively relevant cloud parameters are assessed using radiative transfer calculations and reveal a complex phase dependence of the BTDs, where the dominant link between BTDs and phase is through the cloud top temperature. This analysis helps to understand the potential and limitations of BTDs in phase retrievals. Using these findings, the new PRObabilistic cloud top Phase retrieval for SEVIRI (ProPS) is developed. It employs a probabilistic Bayesian approach for cloud and phase detection based on collocated DARDAR-SEVIRI data. ProPS distinguishes between clear sky, optically thin ice, optically thick ice, mixed-phase, supercooled, and warm liquid clouds. The retrieval has a high (>80%) probability of detection for liquid and ice pixels and classifies more than half of the challenging mixed-phase and supercooled clouds correctly. The new method enables the study of the temporal evolution of cloud phases, in particular also mixed-phase and supercooled clouds, which have so far been rarely studied from geostationary satellites. This thesis contributes to the global effort to observe and understand cloud phases in order to improve their representation in numerical models and to constrain the large uncertainties in climate projections.