Sea-ice loss in Eurasian Arctic coast intensifies heavy Meiyu-Baiu rainfall associated with Indian Ocean warming

The Meiyu-Baiu rainfall, a quasi-stationary rain belt spanning June to July over East Asia, significantly impacts the densely populated region from central-eastern China to Japan. Extreme events, as witnessed in 1998 and 2020, lead to disastrous floods, causing substantial economic losses and fatalities. Accurate prediction of such extreme events is crucial for mitigating their devastating effects. Traditionally, heavy Meiyu-Baiu rainfall has been associated with El Niño and subsequent IO warming in spring and summer. IO warming creates an anomalous anticyclone over the Indo-Northwest Pacific, bolstering moisture supply and upward air movement over the Meiyu-Baiu region via strengthened low-level southwesterlies. However, relying solely on low-latitude SST anomalies for prediction remains insufficient, as exemplified by the inaccurate prediction of the July 2020 rainfall despite accurate June prediction. Therefore, the identification of other pivotal factors, especially those interacting with tropical influences, is vital. Rapid Arctic sea-ice loss and amplified warming have demonstrably impacted mid- and high-latitude atmospheric circulation. While observational studies hint at a link between Siberian coastal sea-ice loss and the 2020 Meiyu-Baiu rainfall, the relationship's consistency and underlying mechanisms remain unclear. This study proposes that Arctic sea ice significantly influences East Asian summer rainfall only in conjunction with IO SST conditions, a hypothesis supported by previous findings showing improved simulation of the 2020 event when Arctic sea-ice concentration (SIC) anomalies are included.

Existing research extensively links heavy Meiyu-Baiu rainfall to El Niño and associated Indian Ocean warming. This warming strengthens southwesterlies, enhancing moisture and uplift over the affected region. However, these models often lack predictive accuracy, particularly for extreme events like the July 2020 rainfall. Other studies explored the connection between Arctic sea ice loss and East Asian summer rainfall, suggesting a potential influence, but the relationship's temporal consistency and underlying mechanisms remain debated. The limited length of reliable observational data and variations across studies contribute to this uncertainty. The current research builds upon these findings by investigating the synergistic effect of Arctic sea-ice loss and Indian Ocean conditions on Meiyu-Baiu rainfall.

This study leverages a 100-member large ensemble of historical simulations from the Community Earth System Model v2 (CESM2), referred to as CESM2LE, covering 1950-2014. The CESM2LE accurately simulates Meiyu-Baiu rainfall characteristics. To isolate the internal variability from the forced warming signal, the ensemble mean (representing the forced signal) is removed from individual runs. Years with May IO SST outside the ±1 standard deviation (std) range are categorized as warm IO (WIO) or cold IO (CIO) cases. Similarly, years with May Kara Sea (KS) SIE outside the ±1 std range are defined as high SIE (HSI) or low SIE (LSI) cases. Combining these categories generates WIO-LSI, WIO-noLSI, LSI-noWIO, CIO-HSI, and CIO-noHSI cases. The analysis focuses on the interannual variability by removing the long-term trends from the observational data, accounting for the forced response to global warming. The lack of significant correlation between May KS SIE and IO SST in individual CESM2 simulations, as well as observations, confirms KS SIE’s potential as an independent predictor. The Community Atmosphere Model version 6 (CAM6) is used for further experiments to isolate the atmospheric response, incorporating SIC and SST anomalies from CESM2LE simulations, The key indices used include: Niño3.4 Index (measuring ENSO), a hybrid two-dimensional blocking index (identifying atmospheric blocking events), and decompositions of moisture convergence and temperature advection to understand underlying physical processes. A statistical model is developed using multiple linear regression on CESM2LE data to predict July precipitation based on May IO SST and KS SIE. This model is then validated using observational data.

The CESM2LE simulations reveal that in years with warm IO conditions, low KS SIE in May significantly intensifies Meiyu-Baiu rainfall. The rainfall in WIO-LSI cases is ~50% greater than in WIO-noLSI cases in July. The increased rainfall is primarily attributed to a substantial (~90%) increase in convective precipitation, particularly around the middle-lower reaches of the Yangtze River. This enhancement also prolongs the Meiyu-Baiu season into July, increasing the risk of flash floods. Analysis of daily precipitation reveals a doubled likelihood of extremely heavy rainfall events comparable to or exceeding the 2020 event in WIO-LSI cases. The increased convective activity is linked to enhanced convective available potential energy (CAPE), resulting from mid-upper tropospheric cooling and increased near-surface moisture in the western Meiyu-Baiu region, and abundant moisture throughout the troposphere in the eastern region. This dynamic is driven by a quasi-barotropic dipole pattern in geopotential height, featuring a high-pressure anomaly over East Siberia and a low-pressure anomaly over the northern flank of the Meiyu-Baiu region. CAM6 simulations, forced with observed SIC and SST anomalies, replicate this dipole pattern, demonstrating that sea-ice loss drives this atmospheric response. The statistical model incorporating IO SST and KS SIE significantly improves the prediction of July Meiyu-Baiu rainfall, particularly in years with low sea ice, like 1997, 2007, 2011, 2012, and 2015. This improvement is evident in reduced prediction bias across the Meiyu-Baiu region and major cities in East China, southern Japan, and South Korea. For instance, the prediction bias for Wuhan and Shanghai was reduced by ~50% for July 2020.

The study's findings directly address the research question by demonstrating the significant synergistic effect of IO warming and Arctic sea-ice loss on extreme Meiyu-Baiu rainfall. The identification of low KS SIE in May as a crucial predictor, especially when IO warming is present, significantly advances subseasonal prediction capabilities. The increased understanding of the underlying physical mechanisms—the quasi-barotropic dipole pattern driving moisture convergence and mid-upper tropospheric cooling—provides a robust explanation for the observed rainfall enhancement. The improved predictive accuracy of the statistical model showcases the practical implications of incorporating Arctic sea ice data into seasonal forecasts. This has significant implications for disaster preparedness and mitigation in East Asia, offering a lead time of one to two months for preventive actions.

This research highlights the critical role of Arctic sea-ice loss in intensifying IO warming-induced heavy Meiyu-Baiu rainfall. The findings demonstrate the significant predictive improvement gained by incorporating Arctic sea ice data into forecasting models. Future research should focus on refining the statistical model, exploring additional influencing factors, and improving the predictive skill for Arctic sea ice extent to enhance the forecast lead time. Investigating the precursors of early Arctic sea-ice retreat could further extend the predictive horizon.

The quantitative results may be model-dependent, as the CESM2LE simulations may not fully capture the nuances of daily extreme rainfall. The statistical model's explanatory power remains limited, suggesting additional factors influence Meiyu-Baiu rainfall. The current model's limited predictive skill for Arctic sea ice extent restricts the forecast lead time. Future work could employ more sophisticated statistical methods and investigate the causes of early sea-ice retreat.