Rainfall events once catalogued as "once-in-a-century" occurrences in New Zealand are now perceived as the new normal, according to a recent study that leveraged artificial intelligence to analyze the impact of global warming. Researchers combined advanced predictive models with decades of precipitation measurements to quantify how rising temperatures could reshape intense rainfall episodes. The result: a marked intensification of the most violent downpours, accompanied by forecast uncertainty that compels a rethinking of hydrological risk management.
What the study reveals: intensification of extreme rainfall and increased frequency
Simulations from the machine learning model indicate that the most intense rainfall events, formerly classified as "one-in-a-hundred-year" occurrences, will gain in average intensity, with precipitation peaks far exceeding historical thresholds. This trend is observed especially in the South Island's watersheds, where thunderstorms intensify under the effect of a warmer and more humid atmosphere. Although the study does not specify an exact percentage, it highlights a clear trend toward a greater probability of seeing these extremes repeat at shorter time intervals.
Furthermore, researchers found that the spatial distribution of violent episodes extends toward coastal areas, thereby increasing the risk of flash flooding in communities not historically exposed to such floods. Forecast uncertainty, linked to natural climate variability, remains high, but AI models reduce this gap by assimilating more atmospheric data, improving the ability to anticipate critical episodes.