Developing a Framework for Evaluating Sources of Predictability for Extreme Events on Subseasonal Timescales in Southeast Asia & South America

Abstract

Rising temperatures due to climate change are leading to an increase in extreme weather events, posing significant risks to vulnerable regions worldwide. These areas are already experiencing extreme precipitation, droughts, and floods, which threaten climate security and endanger lives and property. Subseasonal prediction can provide more timely information for effective climate adaptation strategies than short-term forecast with lead times of 1 - 2 weeks. This study aims to develop a framework, or generalizable methodology, for analyzing past extreme weather events and their predictability to improve subseasonal forecasts and make them more useful for organizations involved in disaster preparedness and climate security. These include international and national agencies such as the Global Water Security Center (GWSC), emergency management authorities, and local governments, organizations, and individuals in climate-sensitive regions. In this work, recent extreme weather events are analyzed, including extreme precipitation events that caused extensive flooding in Pakistan during 2022 and Peru in 2017 and 2023. Using ECMWF Reanalysis v5 (ERA5) and observational datasets, such as Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) data, potential drivers and mechanisms behind these events are examined. NCAR Community Earth System Model Version 2 (CESM2) skill is used to understand how far in advance potentially useful information could have been provided to organizations like the GWSC to be translated to impacted countries. Further, a unique set of initialized prediction experiments from NCAR CESM2 with climatological initial conditions for different combinations of earth system components (e.g., land, ocean, and atmosphere) are used to identify which components contribute most to the occurrence and prediction of these events. These experiments are used as a springboard to analyze the drivers and mechanisms of these events. For the 2022 precipitation in Pakistan, atmospheric initial conditions contributed more to forecast skill in early July and August, while oceanic initial conditions had a greater influence in late July. These results provided a launching point for assessing potential drivers. Our findings indicate that anomalous easterly wind south of the Tibetan Plateau and an enhanced Somali Jet, both influenced by sea surface temperatures, were primary contributors to the 2022 Pakistan floods. Moisture flux analysis indicated that northward propagating monsoon surges were not primary drivers. In Peru, oceanic initial condition contribution to skill dominated supporting the role of Coastal El Niño as a potential driver of the precipitation for both 2017 and 2023. Predictive skill was limited for the 2017 Peru and 2022 Pakistan events, but skill improved significantly for the 2023 Peru event, with forecasts providing useful precipitation information up to 3 - 4 weeks in advance. These findings demonstrate the potential of targeted subseasonal forecasts to inform risk reduction efforts across diverse geographic regions.