Mechanisms Contributing to the Heavy Rainfall Associated with a Mei-Yu Front near Taiwan

On 2-3 June 2017, a southward-propagating meiyu front produced torrential rainfall that flooded northern Taiwan. Over the two-day period, rainfall totals exceeding 200 mm were widespread across the island, several locations in New Taipei City reported accumulations over 400 mm, and a maximum event total of 654.5 mm was observed in the Sanzhi district in New Taipei City. The meiyu front responsible for the flooding had a well-defined cloud shield that extended from north of Hong Kong to the easternmost longitude of the Japanese archipelago. In order to study the storm-scale processes responsible for heavy rainfall near the Taiwan Strait, we examine a high-resolution numerical simulation generated by the Weather Research and Forecasting (WRF) model. Overall, the model reasonably recreates the spatial pattern of total rainfall, although the maximum in the northernmost districts of Taiwan is displaced to the southwest and the heavy rainfall over the Central Mountain Range occurs is displaced to the south. Despite these issues, our goals are to evaluate the mechanisms associated with heavy rainfall and assess whether the mechanisms represented by the model are realistic; the simulation is considered to be accurate for these purposes.

To study how the precipitation processes evolve through time, we use a clustering technique to identify the front at each hour of the simulation based on dense clusters of points where convergence and virtual potential temperature exceed certain thresholds. Although an imperfect technique, the general front is trackable through time, enabling statistics of the surrounding environmental and microphysical characteristics. In terms of dynamical processes, there is a slight relationship between the low-level shear strength and the hourly rain accumulations behind the front, but the relationship with the cold pool propagation speed is weak. We hypothesize that this discrepancy occurs since the propagation of the front is not solely controlled by thunderstorm-cold pool dynamics. At short time scales, high graupel mixing ratios are often associated with high rain mixing ratios. This relationship is somewhat problematic given that overestimation of graupel by microphysical schemes is a known problem in tropical precipitation modeling. On longer time scales and in terms of the rainfall distribution, low-level integrated moisture flux exhibits a strong correlation with the total hourly rain accumulations near the front. Finally, we compare the simulated microphysical structure with operational polarimetric radar data; this comparison allows us to assess the realism of the microphysical structures from the numerical simulation.