Real-Time Estimation of Loss of Life from Urban Dam Breach Based on Population Heat Maps and Land Use Data

Abstract

Accurate estimation of loss of life caused by dam breaches is critical for reservoir risk management and emergency planning. Traditional methods rely on static population data with poor timeliness and low spatial resolution. This study proposes a real-time estimation method by integrating population heat maps with land use data. A temporal variable is introduced to distinguish newly inundated populations from those already counted in previous time steps, thereby avoiding double counting during flood routing, while differences in public understanding of flood severity are considered to evaluate loss of life under various warning conditions. Taking the Shiyan Reservoir in Shenzhen as a case study, the BREACH model and HEC-RAS 2D are employed to simulate dam breach and flood routing. Results show that the final inundation area reaches 16.99 km2, with residential areas accounting for 97.13%. The total at-risk population after land use correction is 200,963, which is 32.7% higher than the uncorrected estimate of 151,482. Loss of life estimates are highly sensitive to warning time and public understanding, reaching 129,691 under the most adverse conditions (no warning, vague understanding) but dropping to 20 when warning time exceeds 1 h with precise understanding. The proposed method provides an effective tool for real-time urban dam breach risk assessment.