The internal erosion effect causes fine particles in the soil to move through seepage, and the loss of these fine particles leads to changes in porosity, which in turn affects the soil’s hydraulic properties and mechanical performance, posing a threat to the safety of dam and levee engineering. To understand the formation and development of internal erosion under reverse seepage, a simulation test device for internal erosion was designed, and experiments were conducted on three granite residual soil samples with identical soil properties under different water flow speeds (25 L/H, 50 L/H, and 100 L/H). By comparing and analyzing the wetting front, the amount of internal erosion, and the water content, the influence of water flow speed on reverse seepage internal erosion was studied. The results show that under reverse internal erosion, as the water flow speed increases, the internal erosion rate accelerates, as evidenced by the faster advancement of the wetting front and the increase in cumulative internal erosion. As internal erosion develops, the fine particle accumulation curve enters a stable phase. After the soil’s water content reaches its peak, it slightly decreases and then remains relatively stable. Fluctuations in the soil water content occur due to the formation of preferential internal erosion channels or the redeposition of fine particles. The soil particle movement, fine particle loss, and redeposition caused by internal erosion create an internal erosion channel that narrows from the inlet to the outlet.

The flood losses caused by storm surges are very severe in coastal cities, which highlights the importance of quantitative assessment of disaster risks. This study estimates the potential life and vehicle damage in the coastal city of Macau based on flood scenarios and storm surge probabilities. The hydrodynamic model considering surface flow and drainage system is applied to simulate floods caused by storm surges. Based on the flood results of the model and the vulnerability, a distribution map of mortality and car damage in the study area was obtained. The proportion and total loss of flood life and cars corresponding to different probabilities are calculated. We found that about 20% of the study area has a flood mortality exceeding 0.003 and cars are damaged when the exceedance probability of flood is below 0.005. When the exceedance probability of flood is less than 0.001, it may result in about 5 fatalities and car losses exceeding 13 million. This study provides a valuable reference for improving flood management and mitigation.