Unseen but increasing: Recent changes in risk of extreme precipitation over Southern Africa and Southeast Asia
This research assesses extreme precipitation risks in Southern Africa and Southeast Asia using the UNSEEN approach. Findings reveal that rainfall risk has doubled in major cities since 1981. It identifies “sitting duck” regions facing increasing risks of unprecedented, record-breaking events that exceed current disaster preparedness and future climate projections.
Please login or join for free to read more.
OVERVIEW
Introduction
In recent decades, there has been a global increase in extreme weather. In Southeast Asia, heat extremes and heavy precipitation have risen, while Southern Africa has experienced more frequent extreme heat, heavy precipitation, and droughts. These extremes cause major infrastructure damage and disease outbreaks, including dengue fever and leptospirosis in Southeast Asia and cholera in Mozambique. This study examines historical changes in extreme rainfall, focusing on the Philippines and Mozambique as examples of coastal countries with high risk. Recent events, such as the washing out of bridges in Mozambique due to Tropical Cyclone Jude in March 2025 and Tropical Cyclone Trami displacing almost 1 million people in the Philippines, highlight these growing concerns.
Observational data of extreme weather
Assessing trends in extreme weather requires examining historical records. However, Southern Africa and Southeast Asia have limited station data and relatively short observed records compared to regions like the North Atlantic. This study uses Climate Hazards Group InfraRed Precipitation with Station data v2.0 (CHIRPS) as a proxy for observational data. While CHIRPS has shown a negative bias in Mozambique, Madagascar, and the Philippines, it has been found to distinguish accurately between days with and without rainfall and has outperformed other observational products during specific cyclone seasons.
Risk of extreme rainfall and future projections
Accurate estimation of extreme rainfall risk must capture both thermodynamical and dynamical changes. A warmer atmosphere holds more water vapour, with precipitation rates from tropical cyclones estimated to increase approximately 7% for every 1°C of global warming. Beyond this, models must account for storm dynamics, such as translation speed and track changes. Slower movement can increase rainfall accumulations and flood risk. In the Philippines, tropical cyclones contribute significantly to rainfall, though only 15% of recorded high precipitation events between 1958 and 2017 were associated with landfalling cyclones.
Breaking records
Changes to the risk of record-breaking events are difficult to detect due to sparse observation networks and physical damage to equipment during extremes. Experience of extreme storms also affects risk perception and preparedness. For instance, the unfamiliarity with super typhoons in Mindanao, Philippines, contributed to the damage caused by Super Typhoon Bopha in 2012. Researchers now use the UNprecedented Simulated Extreme ENsemble (UNSEEN) approach, which utilizes large ensembles of climate simulations to identify “soft records”—places where observed records are not very extreme relative to the local distribution.
Methods
To assess risk, the study compares extreme value analysis of CHIRPS observational data against the synthetic UNSEEN dataset. The UNSEEN ensemble is built using archived historical forecasts from the SEAS5 seasonal forecasting system, providing a larger sample of plausible extreme events. The data is regridded to a 1-degree resolution. Fidelity tests, such as the Kolmogorov–Smirnov test, are employed to ensure the SEAS5 model accurately captures local climate statistics. This test excluded 42% of gridbox-month combinations in Southern Africa, primarily in the dry season, but only 15% in the Philippines.
Results
The study finds that the risk of extreme rainfall changed substantially between 1981 and 2024. In many locations, a non-stationary model provides a better fit than a stationary one. Southern Africa shows a strong trend of increasing extreme precipitation during the rainy season (November to March). In Mozambique, risk increases are most pronounced in the northern part during January and February, when events that were previously 1-in-100-year occurrences now happen every 75 years or less. Risk has doubled in some months for major centres including Phnom Penh, Vientiane, Bangkok, Hanoi, Maseru, Lilongwe, and Lusaka. Conversely, many parts of Southern Africa show a decrease in extreme precipitation likelihood during the dry season (May to September).
Comparison to future projections and “sitting ducks”
The historical UNSEEN changes generally align with CMIP6 future projections. However, in regions such as Northern Mozambique and the Southern Philippines, UNSEEN estimates of increased risk since 2005 already exceed CMIP6 projections for 2021–2040. The study identifies “sitting ducks”—locations with increasing risk but no recent extreme experience. Much of Mozambique, the Philippines, and Laos are considered sitting ducks for extreme precipitation in at least one month of the year. For example, in Metro Manila, what used to be a 1-in-100-year event is now a 1-in-77 year event.
Discussion and conclusion
Many places in Southern Africa and Southeast Asia are “sitting ducks” where risks are increasing in the background without recent rare events to drive preparedness. UNSEEN models produce plausible extreme rainfall events several times larger than any observed since 1981; in some parts of Mozambique, simulated maximum rainfall was three times larger than the historical record. The lack of anticipation for such events can be fatal, as seen in Naga City in October 2024, where rainfall reached 740 mm within 24 hours, surpassing the 100-year return period record of 507.4 mm. Disaster risk managers should use large ensembles to stress-test health systems and infrastructure design, moving from reactive responses to proactive strategies against record-shattering extremes.