Analysis of Temporal Patterns, Monitoring Techniques, and Risk Evolution for Floods and Cyclones in Sri Lanka (1975–2025): Evidence for Paradigm Shift from Reactive Response to Anticipatory Resilience

Sri Lanka has witnessed a struggle with floods and cyclones, transitioning from occasional emergencies to persistent and growing threats. This study provides the first consolidated, long-term temporal analysis synthesising 50 years (1975–2025) of meteorological trends with the parallel evolution of monitoring technology. Using Mann-Kendall trend analysis and Pettitt change-point detection, statistically significant increases in cyclonic disturbance frequency (1.3 events/year in 1975–1985 to 4.1 events/year in 2015–2025; Sen's slope: +0.73 systems/decade, 95% CI: +0.58 to +0.88 systems/decade, p<0.001) and extreme rainfall intensity (12.2 mm/decade increase in annual maximum 1-day rainfall; 95% CI: 9.4–14.8 mm/decade) were demonstrated. Change-point analysis identified 1998 as a statistically significant inflection year (Pettitt test, p<0.05), after which the cyclonic rainfall contribution increased by 18–25% in the northern and eastern provinces of Sri Lanka. While monitoring capabilities have advanced from rain-gauge dependence (pre-1990) through the satellite era (1990s–2000s) to real-time Earth observation integration (post-2015), a critical institutional gap persists: advanced forecasts fail to trigger effective grassroots actions. This "last-mile" problem is operationally defined as the persistent gap between the forecast lead time (72+ hours by 2025) and the actual evacuation completion times (18–36 hours), quantified through lead-time analysis by comparing forecast windows to evacuation completion times. It revealed no proportional improvement in response efficacy despite a 6-fold increase in forecast lead time (from 12–24 hours in 1975–1990 to 72–168 hours in 2015–2025). An integrated Sri Lanka Multi-Hazard Risk Monitoring and Decision-Support Platform (SL-RISK) that couples technical data integration with community-embedded monitoring, impact-based forecasting, and pre-arranged institutional response mechanisms is proposed, with a phased 18–36-month implementation roadmap. Without addressing this governance-technology mismatch, technological advances remain underutilised, perpetuating cycles of preventable disaster losses among vulnerable populations.