Abstract: The information gathering for disaster relief capabilities in major natural disasters suffers from incomplete elements, low accuracy, and difficulty in application. A dynamic analysis model for regional-scale disaster relief capacity distribution and gaps was developed for two types of natural disasters—floods and earthquakes. Based on survey-collected historical data on typical flood and earthquake disasters, as well as disaster relief efforts across the country, a national sample database of typical natural disaster scenarios and relief efforts was established. Tailored to the distinct characteristics of flood and earthquake disaster relief, various algorithms and spatial overlay analyses were employed to develop assessment methods and algorithmic models for flood disaster relief capacity. Using the sample database, machine learning training was applied to construct a seismic disaster relief capacity assessment algorithm model with reasonable generalization capabilities. The models were validated using three representative historical disasters: the Haihe River Basin "23·7" extreme flood event (specifically the Langouwa Flood Storage and Detention Area), the 2021 Haihe River Basin storm flood (involving the Liuweipo and other flood storage and detention areas), and the 2023 Jishi Mountain "12·18" earthquake. The validation results demonstrated high overall accuracy of the model. This model can provide relatively precise assessments of disaster relief capacity during the initial phases of major natural disasters, and supporting information for management authorities to make rapid decision-making.