Abstract:
This study develops an Artifcial Neural Network (ANN) to classify satellite imagery from Multi-angle Imaging SpectroRadiometer (MISR) in domains of (200 km)2 into four categories of marine low-clouds based on their type of Mesoscale Cellular Convection (MCC). These categories are (i) closedcelled MCC (ii) open-celled MCC (iii) disorganised MCC and (iv) No MCC. These different types of MCC are usefully defined as low-clouds of different morphologies. These classifcations are used to investigate the large-scale meteorological controls on MCC. The large-scale meteorological variables that were used in this study are sea-surface temperature (SST) and Lower- Tropospheric Stability (LTS). Changes in large-scale meteorology are found to impact the occurrence of each MCC type disproportionately. We also investigated relationships between the El Ni~no Southern Oscillation (ENSO) and MCC. MCC is found to be strongly inuenced by the SST anomaly patterns that arise during El Ni~no and LaNi~na. Changes in the coverage of MCC during ENSO phases are found to have significant impacts on the Top-Of-Atmosphere albedo. Classifcations from the ANN are also combined with satellite observations from MISR to develop relationships between cloud morphology, domain albedo, cloud fraction and a cloud heterogeneity. Cloud morphology is found to play an essential role in modulating these relationships. The cloud fraction-albedo relationships are found to be directly a function of cloud morphology. Relationships between domain albedo and cloud heterogeneity are also found to be a function of MCC type. Our results strongly indicate that the albedo has a strong dependence on cloud morphology and cloud heterogeneity. Understanding both the physical properties and the meteorological controls on MCC has important implications for understanding low-cloud behaviour and improving their representation in General Circulation Models.