Large-Scale Drivers of Tropical Extreme Precipitation Events: The Example of French Overseas Territories

Due to their severity and lack of predictability, understanding and forecasting extreme precipitation events (EPEs) is critical for disaster risk reduction. The present work documents the large-scale environment of tropical EPEs based on a 42-year data set combining dense rain-gauge networks that cover several tropical small islands and coastal regions. Approximately 10%-30% of EPEs are associated with a tropical storm or cyclone (TC), except for Reunion, for which its high topography makes it reach 55%. TCs multiply the EPE probability by a factor of 4-15, especially during TCs of category 1 or higher. A composite analysis demonstrates that the remaining large part of EPEs occurs within large-scale and strong moist, convective, and cyclonic wind anomalies resulting from the superimposition of intraseasonal, seasonal-to-annual, and interannual timescales. These intense anomalies come essentially from intraseasonal variability, and lower frequencies improve the effect of intraseasonal events in creating a favorable environment for EPEs. Floods and landslides, mainly caused by extreme precipitation events (EPEs), are severe disasters that affect populations and economies. However, their prediction, especially in tropical areas, is challenging. Here, we study the main atmospheric configurations leading to EPEs using rain-gauge data densely sampling several tropical small islands and coastal regions, covering the period 1979-2021. In areas prone to tropical storms and cyclones (TCs), these systems account for 10%-30% of EPEs, except for Reunion, whose high topography increases it up to 55%. TC-related EPEs are also more intense than non-TC-related EPEs. TCs multiply the EPE probability of occurrence by 4-15 times, depending on the region and the TC category. Non-TC-related EPEs are shown to occur within specific large-scale weather patterns in which moisture is highly increased, and convective storms are more active. These patterns involve phenomena occurring at intraseasonal (2-90-day periods), seasonal-to-annual (91-365-day periods), and interannual (periods above one year) timescales. The intraseasonal variability helps the most to develop this environment favorable to the occurrence of EPEs. Slower signals build a background on which intraseasonal variability can have more effects. The real-time monitoring of these different features should improve EPE forecasting capabilities and deliver more early warnings. Over most of the studied tropical small islands, only 10%-30% of extreme precipitation events occur near a tropical storm or cyclone Tropical storms and cyclones favor heavier rainfall and raise the probability of occurrence of extreme precipitation by a factor of 4-15 The other largest part of events occur in a large-scale, intense, moist, and convective anomaly driven mainly by the intraseasonal timescale