Recently, extreme precipitation has increased globally. Waterlogging disasters caused by rainfall have endangered people's lives and caused property losses. As the cell of the city, the community is the fundamental unit to resiliently withstand disasters. However, stress testing in the community resilience field is still in its infancy, and little research exists on relevant theoretical and technical frameworks. Therefore, this paper uses a rainstorm waterlogging disaster as an example to study the community resilience stress test method and provides application cases.

The community resilience stress test stimulates and assesses community resilience in response to various emergency events. Herein, the resilience curve method is used to calculate community resilience. This paper presents a specific community resilience stress test method for rainstorm waterlogging disasters. First, using the historical rainstorm disaster data and rainstorm intensity formula, 12 extreme rainfall scenarios were designed according to the 2 dimensions of hourly rain intensity and rainfall duration, covering 30-200 mm hourly rain intensity. Second, based on InfoWorks Integrated Catchment Modeling, this paper constructs a community rainstorm waterlogging hydrodynamic model. This study conducted the community rainstorm waterlogging measurement experiment in the rainy season. The monitoring data obtained are used for parameter calibration and validation of the hydrodynamic model. Then, this paper presented a resilience evaluation method focusing on engineering resilience. The system performance of community rainstorm waterlogging is defined by the proportion of inundated areas. The system performance of a drainage network is defined by the fullness of the drainage network. Community waterlogging resilience was calculated using these two types of system performance. Resilience is expressed using the area of the concave portion of the system performance curve of community rainstorm waterlogging. The termination time of the integral is calculated as the time when the system performance of the drainage network returns to 1.

Using the community J in Beijing as an example, this paper conducted a stress test on the community's waterlogging resilience under 12 different extreme rainfall scenarios, based on the results of hydrodynamic simulation. The results show that community resilience is less affected by rainfall duration and positively correlated with hourly rainfall intensity under rainstorm waterlogging disasters. Under the extreme rainfall scenario of 200 mm/h, about 44% of the community was flooded, and the maximum water depth was nearly 1 m. About 95% of drainage pipes are overloaded. It takes 5.7 hours to fully restore the drainage capacity of the network. Waterlogging spots of varying severity in this community are observed. This paper provides targeted suggestions on how to improve community resilience under rainstorm waterlogging disasters for five main waterlogging-prone spots.

This paper proposes a stress test method for community resilience to waterlogging and analyzes the evolution process of resilience and risk tolerance of community J from two perspectives: drainage capacity of pipe network and inundated area of the community. This method provides a quantitative assessment of community resilience. The test results can be used for monitoring and investigating rainstorm waterlogging risk in community institutions and government departments. These results are conducive to preventing and resolving disaster risks in advance.

Rapid prediction of rainstorm waterlogging is crucial for disaster prevention and reduction. However, the traditional numerical models for simulating and predicting large-scale and complex subsurface conditions are complicated and time-consuming; moreover, the time-efficiency requirement of rainstorm waterlogging prediction is difficult to meet. To address these shortages of the numerical models, this study constructs a spatiotemporal prediction model of urban rainstorm waterlogging based on machine learning methods to rapidly predict waterlogging extent and water depth changes.

This study constructs a rapid prediction model of urban rainstorm waterlogging based on a hydrodynamics model and machine learning algorithms. First, a hydrodynamic model is constructed based on InfoWorks integrated catchment management (InfoWorks ICM) for rainstorm waterlogging in the study area with the parameter rate determination and model validation to realize the high-precision simulation of urban rainstorm waterlogging. On this basis, a rainfall scenario-driven hydraulics model is designed to further obtain rainstorm waterlogging simulation results. These results are used as the base dataset for machine learning. Second, the spatial characteristics data of rainstorm waterlogging are obtained from three aspects: rainfall situation, subsurface information, and the drainage capacity of the pipe network, which, together with the grid simulation results, comprise the dataset. The spatial prediction models are based on random forest, extreme gradient boosting (XGBoost), and K-nearest neighbor algorithms. Finally, the simulation results of waterlogging points are used to generate rainstorm waterlogging time series data. The rainfall, cumulative rainfall, and water depth of the first four moments (every 5 min) are used as the input for a long short-term memory (LSTM) neural network to predict the present water depth of the flooding point. The two models collaborate to achieve rapid spatial and temporal predictions of urban rainstorm waterlogging.

For spatial predictions, the random forest model has the best fitting performance regarding evaluation indexes such as the mean square error, the mean absolute error, and the coefficient of determination (R²). When a rainstorm scenario with an 80-year event and a 2.5 h rainfall calendar prediction set is used, the prediction results concur with the risk map of urban waterlogging in Beijing. Compared with the simulation results of InfoWorks ICM, the prediction accuracy of the predicted inundation extent reaches 99.51%, and the average prediction error of waterlogging depth does not exceed 5.00% by the random forest model. For temporal predictions, the trend of the water depth change of the LSTM neural network model is more consistent with the simulation results of InfoWorks ICM, the R² of four typical inundation points are above 0.900, the average absolute error of water depth prediction at the peak moment is 1.9 cm, and the average relative error is 4.0%.

When addressing sudden rainstorms, the rapid prediction model based on machine learning algorithms built in this study can generate accurate prediction results of flooding extent and water depth in seconds by simply updating the forecast rainfall data in the model input. The model computational speed is greatly improved compared to the hydrodynamics-based numerical model, which can help plan waterlogging mitigation and relief measures.