Characterizing the Concurrent Occurrence of Tornadoes and Flash Floods Across Canada

Compound weather extreme events, such as tornadoes and flash floods, can significantly impact societies and infrastructure systems. Disaster response agencies provide instructions to the exposed communities to retreat to safety specific to the natural hazard. However, the instructions can become confusing if natural hazards demand conflicting responses. This study characterizes the compound tornado and flash flood (TORFF) events to assess and predict the simultaneous occurrence probability of such hazards across Canada in the long term. We quantify dependencies between the tornadoes and flash floods using ground-based and reanalysis datasets. Tornado data are available based on the recorded Fujita rating for each event, and the corresponding wind speed values are determined through a resampling approach. The TORFF events are clustered and the bivariate probability distributions of the resampled windspeed and precipitation are characterized based on Copula. The corresponding individual and joint return levels are investigated under different scenarios (AND, OR, and conditional) across Canada. Results show positive dependencies between resampled windspeed and associated precipitation in Saskatchewan, followed by Alberta, Manitoba, Ontario, and Quebec (least dependency) regions. Higher dependencies between tornadoes and flash floods over regions such as Saskatchewan suggest that analyzing these events in isolation can underestimate the associated risks. Higher precipitation is also expected during extreme wind speed, as observed in the conditional assessment of precipitation given windspeed. This study provides insight for more realistic recurrence interval estimation for tornadoes and flash floods to aid in the evacuation decision-making process