Summer rains on the Yangtsey: following the evolution month by month of the oceans in spring improves the seasonal forecast

An index learned by total graph variation, combined with the humidity of Indochina soils, predicts summer rains in the Yangtse basin better than the El Niño indicators.

Predicting summer rainfall in the Yangtze River basin months in advance remains difficult because the ocean signals that predict them do not behave in the same way from year to year. A team shows that taking into account changes in sea surface temperatures over the spring, rather than an average, and soil conditions significantly improves predictability.

A clue learned about three ocean basins

The method applies the total graph variation to surface temperatures from March to May in the western Pacific, El Niño3.4, and the tropical Indian Ocean. It assigns each grid point and month a spatially consistent coefficient, and then condenses them into a single index. The learned patterns reflect the decline of El Niño between April and May and the inverse adjustment of the Indian Ocean, which corresponds to the known coupled Indo-Pacific variability.

Over the period 2008-2022, this index predicts better than approaches based on Niño3.4 or on a parsimonious regression. Adding the interaction between soil moisture on the Indochinese peninsula in May and the temperature of the Indian Ocean further improves the score and corrects for several previously underestimated rainy summers. Dry soil in Indochina associated with a warm Indian Ocean coincides with a more pronounced westward extension of the Pacific subtropical high.