[A-AS03] Large-scale moisture and organized cloud systems

[AAS03-P03] Classification of Quasi-Seasonal Synoptic Fields Using Self-Organizing Map and Occurrence Trend of Line-shaped Rainbands

*Yuta Ohya¹, Tomohito Yamada² (1.Hokkaido University Graduate School of Engineering, 2.Hokkaido University)

Keywords:Meso-scale convective systems, Sub-seasonal

In recent years, Line-shaped rainbands (hereinafter referred to as LRB) have occurred frequently in Japan. LRB is precipitation systems with meso-scale convective systems in which cumulus clouds are continuously distributed in a linear pattern. In Japan, Hokkaido is considered the region where annual rainfall is relatively low compared to that in Kyushu, and other western regions. Extreme rainfall events such as LRB tend to increase in this island, such as the extreme rainfall event in 2010, the frequency of LRB is extremely high in this island. In such years, the sea surface temperature around Hokkaido is higher than climatology, and the positive anomaly associated with the active Pacific high pressure system, brings a huge water vapor supply to the region in sub-seasonally.
Considering the aforementioned information, sub-seasonal meteorological fields were classified by applying machine learning to reanalysis data and large ensemble data in this study. The method is Self-Organizing Map, which has an advantage in mapping multidimensional vectors to two-dimensional vectors.
Meteorological fields similar to those in multiple LRB years would be identified from these classifications. The results indicated that in those clusters, the number of LRB occurrences in the same area was relatively high. In order to take advantage of the application to large ensemble data, the statistical advantages of the classification will be explained in the further study.