Extreme Precipitation Events in Peru: A Data-Driven Classification and Large-Scale Ocean Controls over the past four decades

Extreme Precipitation Events (EPEs) in Peru have major socio-environmental consequences, driving floods and landslides and causing serious damage to infrastructure and food security. To better identify and interpret these events despite the lack of a universal definition of “extreme” rainfall, the study proposes a data-driven framework that (1) fits probability distributions to monthly precipitation anomalies from the PISCO dataset (1981–2025) at each grid point to define local extreme thresholds, (2) measures event intensity with the Relative Exceedance Index (REI), and (3) uses K-means clustering to detect recurrent spatio-temporal regimes of extremes across the country. The results show that this approach successfully captures historically documented episodes of widespread impacts and reveals a small number of coherent, recurring spatial patterns of extreme rainfall. Composite analyses further indicate that these EPE regimes are systematically linked to large-scale climate anomalies: La Niña-related extremes tend to align with central Pacific cooling and North Atlantic warming that favor moisture inflow and heavy rainfall over the northern/central Andes, while El Niño-related extremes are tied to eastern Pacific warming that enhances onshore convection and moisture transport, intensifying coastal and western Amazon rainfall while South Atlantic warming further strengthening Amazon-focused extremes. Overall, this framework not only strengthens the detection and classification of extreme rainfall events, but also provides a robust approach to identifying large-scale oceanic sources of predictability that is crucial for anticipatory planning, risk management, and long-term adaptation strategies.