EGU26-10195, updated on 14 Mar 2026
https://doi.org/10.5194/egusphere-egu26-10195
EGU General Assembly 2026
© Author(s) 2026. This work is distributed under
the Creative Commons Attribution 4.0 License.
Oral | Wednesday, 06 May, 14:33–14:36 (CEST)
 
vPoster spot 3
Poster | Wednesday, 06 May, 16:15–18:00 (CEST), Display time Wednesday, 06 May, 14:00–18:00
 
vPoster Discussion, vP.56
Warning is not enough: time delays and spatial inequalities in household-scale cyclone evacuation in coastal Bangladesh
Md Rajibul Islam, Md Hasanur Rahman, Farzana Ahmed Ahmed, and Dr. Mashfiqus Salehin
Md Rajibul Islam et al.
  • Bangladesh University of Engineering and Technology (BUET), Institute of Water and Flood Management (IWFM), Dhaka, Bangladesh (rajib.reach.bd@gmail.com)

Early warning systems are fundamental to cyclone risk reduction, yet evacuation outcomes depend on whether warnings trigger timely household action and whether households can physically reach shelters. This study quantifies evacuation thresholds, time-to-action, and mobility constraints using georeferenced survey data from 1,126 households across two cyclone-prone coastal unions in Bangladesh. Using household-level GPS data, we measured distance to the nearest cyclone shelter for each household and analysed evacuation behaviour across spatial distance thresholds.

Early warning message (EWM) coverage was high, with 93.5% of households reporting receipt of warnings, yet only 80.8% evacuated, indicating a persistent warning–action gap. Logistic regression shows that households receiving EWMs had more than twice the odds of evacuation (OR = 2.13, 95% CI: 1.27–3.55, p < 0.01), although evacuation likelihood varied significantly by distance to shelters, and road conditions. Distance to shelters and road conditions were also significantly associated with evacuation outcomes (p < 0.001).

Time-to-action analysis indicates delayed mobilisation after warnings: only 39.8% of households began preparation within 1 hour, and 9.3% delayed action beyond 6 hours. Distance and road conditions compounded these delays: evacuation times rose sharply beyond 1 km and were significantly longer where roads were reported poor or waterlogged during the cyclone, suggesting that delayed mobilisation increases exposure to peak travel constraints.

Spatial constraints also explain non-evacuation among warned households. Among households (18%) that received warnings but did not evacuate, the dominant barriers were distance from shelters (50.0%), shelter overcrowding and lack of privacy or maternal facilities (48.9%), and lack of transportation (45.7%), alongside caregiving and health-related constraints. Only 2.4% cited lack of knowledge about shelter locations, indicating that non-evacuation reflects spatial and mobility exclusion rather than information failure.

These findings demonstrate that cyclone evacuation is a threshold-based and constrained mobility process, where warnings increase evacuation odds but do not guarantee timely action for households facing greater distance, degraded road conditions, and care burdens. Strengthening anticipatory action therefore requires addressing spatial inequalities in last-mile accessibility, reducing response delays, and improving shelter suitability for households with health and caregiving needs in high-risk coastal settings.

 

How to cite: Islam, M. R., Rahman, M. H., Ahmed, F. A., and Salehin, Dr. M.: Warning is not enough: time delays and spatial inequalities in household-scale cyclone evacuation in coastal Bangladesh, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10195, https://doi.org/10.5194/egusphere-egu26-10195, 2026.