Potential Seasonal Predictability of the Risk of Local Rainfall Extremes Estimated Using High‐Resolution Large Ensemble Simulations. Issue 24 (14th December 2021)

• Record Type: Journal Article
• Authors: Imada, Y.; Kawase, H.
• Is Part Of: Geophysical research letters. Volume 48:Issue 24(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-12-14
• Subjects: seasonal prediction -- local extreme rainfall -- GCM -- RCM -- large ensemble simulation; Geophysics -- Periodicals; Planets -- Periodicals; Lunar geology -- Periodicals; 550
• DOI: 10.1029/2021GL096236
• Languages: English
• ISSNs: 0094-8276

Abstract

Abstract: Significant progress has been made in seasonal climate predictions based on general circulation models (GCMs). However, seasonal prediction of local heavy rain remains challenging due to limited resolution. A small signal‐to‐noise ratio also prevents meaningful predictions. In this study, 100‐member large‐ensemble sets of GCM simulations and high‐resolution downscaled products with a regional climate model were used to identify the unknown source of seasonal predictability for the risk of local heavy rainfall around Japan and Taiwan. The detected predictable signals depended on the location and the season. Highly predictable signals were found for events caused by typhoons and the stationary front. Further analyses of the large‐scale backgrounds simulated by the parent‐GCM revealed that each source of predictability comes from various flavors of El Niño/Southern Oscillation and anomalies in the Indian Ocean. The approach of this study potentially provides an opportunity to identify unknown predictable regional signals all over the world. Plain Language Summary: Recently, significant progress has been made in seasonal‐scale climate predictions (forecasts of up to months) based on a global climate model. However, the spatial resolution of a commonly used global climate model is insufficient for simulating local heavy rainfall events. In addition, there are insufficient samples for analysis because of the rare occurrence of extreme events. Therefore, the seasonal prediction of local heavy rainfall remains challenging. In this study, we resolved these shortcomings by increasing the number of samples through repeated numerical simulations using a high‐resolution climate model, and examined whether this could improve predictions. Our analysis covers the regions of Japan, the Korean Peninsula, and Taiwan. We found that heavy rainfall frequency is highly predictable in Taiwan during the typhoon season and in western Japan during the rainy season. This is because each heavy rainfall event is strongly linked to different sea surface temperature anomalies in tropical oceans that can be predicted with high accuracy using a global climate model. Various phases of El Niño and La Niña and the related Indian Ocean fluctuations are the primary drivers of heavy rainfall in each region. Our approach potentially provides an opportunity to identify unknown predictable regional signals worldwide. Key Points: This study employs high‐resolution large‐ensemble simulations to overcome shortcomings of seasonal prediction of heavy rain events Large‐ensemble simulations with 20‐km grid spacing are essential to obtain high predictability of heavy rain frequency in Taiwan and Japan The sources of the predictability come from sea surface temperature anomalies in the tropical Pacific and Indian Oceans