Search for a command to run...
The growing risk of flash floods, driven by more intense sub-daily rainfall extremes under climate change, poses a significant threat to human safety and urban infrastructure. This study highlights the need for revision of rainfall warning thresholds in Croatia where existing criteria are based on limited historical data. A consistent climatological understanding of sub-daily rainfall extremes is therefore a crucial first step toward developing impact-based warning system. The study examines the climatology of 3-, 6-, 12-, and 24-hour rainfall extremes using data from 54 stations across Croatia for the period 1961–2020. The Generalized Extreme Value (GEV) distribution was applied to estimate return levels for multiple return periods, using both stationary and non-stationary models (with time-varying location and scale parameters) to assess potential climate change effects on exceedance probabilities. The results reveal a distinct spatial pattern: the lowest rainfall extremes occur in eastern continental areas, while the Adriatic region, including the coast, hinterlands, and highlands, experiences the highest intensities. Seasonally, rainfall maxima occur mostly in summer over the mainland and in autumn along the coast, consistent with cyclone tracks and regional climate drivers. Trends in annual maxima indicate increases at around 20% of stations. However, no statistically significant differences were found between stationary and non-stationary return values. Based on return level estimates, we propose using the 2-, 5-, and 10-year return levels as reference thresholds for moderate, severe, and extreme rainfall warnings, respectively. Nonetheless, continued monitoring is essential to assess whether future changes in extreme rainfall will justify a shift toward non-stationary approaches in operational practice.
Published in: Theoretical and Applied Climatology
Volume 156, Issue 11