Comparing the Characteristics of Modeled and Observed Mesoscale Cellular Convection in Marine Cold Air Outbreaks

During cold-air outbreaks (CAO) when cold air advects from boreal continents or the Arctic sea ice over open water, a well-recognized cloud pattern forms with linearly organized cloud streets and, further downwind, open or closed cellular convection. Despite the ubiquity of this CAO cloud regime over high-latitude oceans, its properties are rather poorly understood. In this study, large eddy simulations of 12 different CAO periods with either open or closed cellular convection are compared to observations of the same periods. These observations were collected as part of the Cold-air Outbreaks in the Marine Boundary Layer Experiment (COMBLE). Vertical transects of equivalent radar reflectivity factor from the Ka-band ARM zenith radar (KAZR) are used to objectively identify individual convective cells. Similarly, individual cells are identified in the model data by emulating the KAZR with a radar simulator, allowing an apples-to-apples comparison of the characteristics of observed and simulated clouds. Overall, the observed and simulated cells compare well in terms of the total number of cells, and the distributions of cloud top height, cloud top temperature, and liquid water path. One exception is a closed cellular CAO period during which the simulated vertical cloud structure differs substantially from the observations. To further investigate the characteristics of all cells, individual cells are classified using the unsupervised machine learning algorithm self-organizing maps. Results from the classification using observational data show that open cells undergo a convective lifecycle with different properties at each lifecycle stage. This lifecycle is also revealed by the classification of the simulated cells.