Improvements of Procedures for Identifying and Classifying Heavy Rainfall Areas of Linear-Stationary Type

Abstract

The procedures for identifying and classifying heavy rainfall areas of linear-stationary type (LS-HRAs) from the distributions of radar/raingauge-analyzed precipitation amounts (RAP) were improved to extract “senjo-kousuitai” with elongated and stagnant characteristics that causes localized heavy rainfall in Japan. The improved procedures were verified based on subjective judgments (‘certain’, ‘doubtful’, and ‘suspect’) whether LS-HRAs possess the characteristics of typical senjo-kousuitai. Criteria for excluding LS-HRAs judged to be ‘suspect’, mainly associated with tropical cyclones (TCs) and non-stationary mesoscale convective systems, were introduced to optimally extract senjo-kousuitai events. The criteria were determined by the distance between LS-HRAs and TC centers (DLT), the maximum total RAP of LS-HRA (total RAP), and the properties of each RAP distribution at every hour during the LS-HRA extraction period, e.g., DLT ≤ 500 km and total RAP < 200 mm. By applying these criteria, 372 senjokousuitai events were extracted from 452 LS-HRAs during the warm seasons (April–November) in 2009−2020, although approximately 35% of the LS-HRAs judged to be ‘suspect’ were not excluded. The criteria associated with TCs excluded them more effectively than the other factors. The improved procedures for extracting senjo-kousuitai events are expected to be used effectively for their statistical analyses.