Objectively Diagnosing Characteristics of Mesoscale Organization from Radar Reflectivity and Ambient Winds

In the classical model of mesoscale convective systems (MCSs), a system generates new convective cells on the down-shear side of its cold pool, with the cells fed at low levels from the front, and the stratiform cloud trailing behind the system in the up-shear direction, where "front" and "behind" typically refer to the system's ground-relative velocity. In this study we present the "MINT" algorithm, which identifies and tracks MCSs in radar reflectivity data, and objectively diagnoses organisational characteristics related to the classical model, namely the offset of stratiform cloud from convective cloud relative to system velocity, the low-level inflow direction, and the shear-relative tilt and propagation directions. When applied to the 15 year radar record covering the Darwin region of northern Australia, the algorithm indicates 65-80% of MCS observations are consistent with the classical model, at least when the four classifications can be made unambiguously. However, these observed characteristics occur almost entirely in the drier phases of the Australian monsoon. During the humid, active monsoon phase, observed characteristics consistent with the classical model are rare, and most systems exhibit non-classical up-shear propagation. The MINT algorithm can also be used for "fuzzy verification”. When applied to operational convection permitting model and radar observations over Northern Australia, MINT indicates simulated MCSs are approximately twice as likely to be oriented parallel to the ambient wind and ambient wind shear than those observed by radar, indicating a bias toward the "training line" systems typically associated with more extreme rainfall. During highly humid active monsoon conditions, simulated convective systems have larger ground-relative speeds than systems observed in radar. Although there is less than 5% difference between the ratios of simulated and observed trailing, leading and parallel stratiform system observations, significant differences exist in other wind-shear-based classifications. For instance, in absolute terms, simulated systems are 10-35% less likely to be up-shear tilted, and 15-30% less likely to be down-shear propagating than observed systems, suggesting errors in simulated cold pool characteristics.