Abstract:Emergency shelters are important for the evacuation, rescue, and resettlement of citizens, and their spatial layout directly affects the ability of a city to resist disasters. Accessibility and the degree of potential crowding are two key indicators relating to the interaction between the size of the residential area and the emergency shelter; therefore, establishing a reasonable method of evaluating these factors is important to provide a reference value for the design and balanced layout of such shelters. In this paper, three parameters are introduced to improve the gravity model:the population-scale influence factor, the service capacity of emergency shelter, and the travel limit distance. The Chengguan district of Lanzhou City is used as an example for evaluating the spatial layout rationality of urban emergency shelters with respect to accessibility and the potential degree of crowding under different travel limit distances. The results show that there is an unbalanced spatial distribution of emergency shelters in the study area, which would not fully meet the evacuation needs of the current population. With short travel limit distances (500 m and 1 000 m), the accessibility for people in most areas is low, and there are significantly different degrees of potential crowding between the southwest and northeast. With a travel limit distance of 2 000 m, accessibility shows an attenuating trend from northeast to southwest, and the degree of potential crowding is reversed, the crowded degree of each emergency shelter is less than 0.4 person/m2. The improved gravity model comprehensively considers the attenuation of distance, the service capacity of emergency shelters, and the population of the residential area, and it can effectively measure the accessibility of emergency shelters and the potential crowding degree; it can thus be used to reliably evaluate the rationality of their spatial layout.