Potential links between large-scale circulation patterns and extreme precipitation events over the Central-Eastern China

Abstract: Catastrophic extremes not only depend on the large-scale atmospheric circulation situations, but also on water vapor transport. In this study, we use an unsupervised neural network algorithm, self-organizing map (SOM) to identify and visualize the large-scale atmospheric circulation patterns (CPs) over the spatial domain of 10°S–70°N and 40°E–170°W, which represented by standardized anomalies of 850 hPa geopotential height. Theil-Sen estimator, Mann-Kendall (MK), and Pettitt test are chosen to investigate the change trends and abrupt points in the time series of extreme precipitation over the Central-Eastern China (CEC) during 1960–2015. Results show that extreme precipitation over the southeast CEC demonstrates a significant positive trend (at 90% significance level). Regarding the average abrupt points of extreme precipitation frequency and its amount are 1988.3 and 1988.7 respectively, the time series are divided into two periods, i.e., 1960 to 1989 and 1990 to 2015. Then, we objectively adopt 5×5 SOM nodes for each period to represent the atmospheric CPs. At first, the simultaneities for extreme precipitation of 228 rain gauges are examined by using event synchronization, and these gauges are separated into four clusters by using the modularity method. Based on the SOM results, by examining the synchronization degree between extreme precipitation occurrences across the four clusters, we found that the patterns characterized by obvious negative anomalies of 850 hPa geopotential height over the Eastern and Southern Asia continent are highly synchronized with extreme precipitation events around the CEC. Moreover, comparing for each cluster the most synchronized CPs to heavy events of the two periods, we found significant changes in the structure and occurrence of CPs driving heavy rainfall events, that might be a consequence of change of the global pole-equator and ocean-land contrast temperature gradients.

Keywords: extreme precipitation, large-scale circulation patterns, self-organizing map, event synchronization, integrated vapor transport