Incorporating Forecast and Public Response Uncertainty in Agency Timing Decisions for Evacuation Statements

The presentation focuses on the interplay between uncertainty in probabilistic forecasts and uncertainty in public response for an impeding impactful flood event. Exaggeration of the frequency and magnitude of impactful events under the current climate variability and change make this analysis important for improved response and adaptation for the forecast and disaster management agencies worldwide.

The theoretical work is based on relationships obtained from surveys of communities in the United States and on established error characterizations of probabilistic forecasts of impactful event occurrence. The focus of the analysis is to estimate the most effective time to issue evacuation statements for a forecast, impeding impactful event. The methodology allows for adaptive estimation in real time. The work is based on the formulation of Georgakakos (2025) with extensions on the characterization of uncertainty.  It combines decision theory solutions embedded within ensemble Monte Carlo simulations to characterize uncertainty.  Time is normalized to allow for a wide range of forecast lead times and event occurrence times from present time.  The following Figure, adopted from the reference mentioned, indicates the component processes considered.

Open Access Reference: https://doi.org/10.1007/s11069-025-07540-5

Initial results highlight the importance of the time to complete the evacuation and the threshold of probability set by the agencies for action on the timing decision for evacuation statement. Higher time specificity of the forecast probability of occurrence of the impeding event allows for longer times from present for issuing the evacuation statement. As the time of event impact nears with non-negligible occurrence probability at the decision time, higher risk aversion (using high percentile of the decision function versus using expected value of that function) shortens the time to issue the evacuation statement.