A Two-Level Early Warning System on Urban Floods Caused by Rainstorm

Abstract

In recent years, the combined effects of rapid urbanization and climate change have led to frequent floods in urban areas. Rainstorm flood risk warning systems play a crucial role in urban flood prevention and mitigation. However, there has been limited research in China on nationwide urban flood risk warning systems based on rainfall predictions. This study constructs a two-level early warning system (EWS) at the national and urban levels using a two-dimensional hydrological–hydrodynamic model considering infiltration and urban drainage standards. A methodology for urban rainstorm flood risk warnings is proposed, leveraging short-term and high-resolution rainfall forecast data to provide flood risk warnings for 231 cities in central and eastern China. Taking Beijing as an example, a refined rainstorm flood warning technique targeting city, district, and street scales is developed. We validated the methodology with monitoring data from the “7.31” rainstorm event in 2023 in Beijing, demonstrating its applicability. It is expected that the findings of this study will serve as a valuable reference for the urban rainstorm flood risk warning system in China.

1. Introduction

In recent years, the dual pressures of global climate change and rapid urbanization have led to an increased frequency of extreme rainfall and urban flood events [1,2,3,4,5]. These events severely disrupt urban functionality and pose significant threats to lives and property [6,7,8,9]. Particularly, from 17 to 23 July 2021, Henan Province experienced severe flooding, resulting in 398 deaths or disappearances and direct economic losses of 120.06 billion yuan [10]. According to the China Bulletin on Flood and Drought Disasters, over the past decade, more than 100 cities in China annually experienced severe inundation or urban flooding.

Amid this backdrop of frequent extreme weather and recurrent urban flooding during heavy rainfall [11,12,13], the ability to anticipate, assess, and mitigate urban flood risks has become a pressing issue. Timely dissemination of early warning information is essential to guide public responses, safeguard lives, and minimize economic losses. In response, several cities in China, including Beijing, Shanghai, Shenzhen, Wuhan, Nanjing, Xi’an, and Kunming, have initiated exploratory efforts in urban flood early warning systems [14,15,16,17,18,19]. For instance, Yang Biao and colleagues proposed an urban flood monitoring and early warning system applied in selected regions of Shantou City [20], while Cao Xuejian and co-authors developed a comprehensive observation-based urban flood risk early warning system, tested in Beijing’s Qinghe River basin [21].

Advances in numerical flood modeling, the Internet of Things, big data, artificial intelligence, and cloud computing have catalyzed the transition from conventional meteorological rainstorm warnings to more targeted urban flood early warning systems [22,23,24,25,26]. Currently, urban flood early warning methods can be categorized into the following three types. 1. Monitoring-based warning: This approach installs monitoring facilities in flood-prone areas to measure inundation depth and extent, enabling real-time urban flood warnings based on monitored data [27,28,29]. 2. Forecast-based warning using rainfall predictions: Early warnings are issued based on the forecasted rainfall intensity and coverage. However, this method operates on a large spatial scale and does not account for the complex urban rainfall-runoff processes. 3. Forecast-based warning using rainstorm-flood models: This method constructs urban rainstorm flood models driven by high-resolution, short-term rainfall forecasts, real-time hydrological and meteorological data, and operational conditions. It predicts surface inundation depth, duration, and extent, providing short-term, high-resolution, and large-scale urban flood warnings [30,31,32,33].

This paradigm shift allows for proactive flood risk management by predicting urban inundation areas before heavy rainfall events, enabling more precise disaster prevention and mitigation strategies. Unlike traditional approaches that rely solely on storm forecast data, urban flood early warning systems forecast inundation distribution before extreme rainfall, issuing risk alerts tailored to specific scenarios.

Despite promising developments, significant challenges remain. Many developed cities in China face high costs, slow computation speeds, and inconsistent warning metrics in implementing refined flood simulation models. Additionally, there is a nationwide lack of systems capable of providing rapid urban flood risk warnings for multiple cities simultaneously. To address these gaps, the Chinese Ministry of Emergency Management (MEM) in 2022 outlined a national-level requirement for rainstorm-induced urban flood early warning services in central and eastern China. Subsequently, the China Institute of Water Resources and Hydropower Research, in collaboration with the National Disaster Reduction Center, undertook the development of an urban rainstorm flood early warning model.

This study addresses the urgent need for rainstorm-induced urban flood risk warnings in central and eastern Chinese cities. By leveraging two-dimensional hydrological–hydrodynamic models and high-performance computing, we propose an early warning system (EWS). Furthermore, we introduce an administrative-level urban flood early warning system based on forecasted rainfall, providing robust technical support for urban flood prevention and mitigation efforts.

2. Methodology

2.1. Structure of the EWS

The EWS establishes flood models to calculate flood risk elements such as inundation depth and duration and classifies urban flood risk warning levels based on predefined indicators. Urban flood risk warning operates at two levels. At the national level, rapid warnings are issued for 231 prefecture-level and above cities across 24 regions, including Guangdong, Fujian, and Zhejiang, as shown in Figure 1. At the urban level, the focus shifts to flood risks in administrative divisions such as districts, counties, and streets.

The EWS (Figure 2) first conducts national-level warnings, issuing warning levels for 231 cities. Subsequently, for cities with yellow or higher warning levels, urban-level risk warnings are carried out, resulting in risk warning levels for three types of administrative divisions: city-level, district-level, and street-level.

2.2. Model Construction

2.2.1. Hydrodynamic Method

The two-dimensional shallow water equations are widely used for modeling hydrodynamic processes dominated by horizontal water flow, such as surface runoff in urban areas, nearshore waves, and tides. The vector form of these equations is as follows:

$${\displaystyle \frac{\partial \mathbf{U}}{\partial \mathrm{t}}}+{\displaystyle \frac{\partial \mathbf{F}}{\partial \mathrm{x}}}+{\displaystyle \frac{\partial \mathbf{G}}{\partial \mathrm{y}}}=\mathbf{S}$$

(1)

In the equations, the variables are defined as follows: t is time, U represents the vector of conserved variables, including water depth (h) and momentum (hu and hv), F and G are the flux vectors along the x- and y-axes in the Cartesian coordinate system, representing momentum transport in their respective directions, and S is the source term vector, which includes factors such as rainfall, infiltration, bed slope, and friction.

$$\mathbf{U}=\left[\begin{array}{c}\mathrm{h}\\ \mathrm{hu}\\ \mathrm{hv}\end{array}\right],\hspace{0.33em}\hspace{0.33em}\mathbf{F}=\left[\begin{array}{c}\mathrm{hu}\\ \mathrm{h}{\mathrm{u}}^2+{\displaystyle \frac{1}{2}}\mathrm{g}{\mathrm{h}}^2\\ \mathrm{huv}\end{array}\right],\hspace{0.33em}\hspace{0.33em}\mathbf{G}=\left[\begin{array}{c}\mathrm{hv}\\ \mathrm{huv}\\ \mathrm{h}{\mathrm{v}}^2+{\displaystyle \frac{1}{2}}\mathrm{g}{\mathrm{h}}^2\end{array}\right]$$

(2)

$$\mathbf{S}=\mathbf{RI}+{\mathbf{S}}_{\mathrm{b}}+{\mathbf{S}}_{\mathrm{f}}=\left[\begin{array}{c}\mathrm{R}+\mathrm{I}\\ 0\\ 0\end{array}\right]+\left[\begin{array}{c}0\\ \begin{array}{c}\text{-}\mathrm{gh}{\displaystyle \frac{\partial \mathrm{z}}{\partial \mathrm{x}}}\\ \text{-}\mathrm{gh}{\displaystyle \frac{\partial \mathrm{z}}{\partial \mathrm{y}}}\end{array}\end{array}\right]+\left[\begin{array}{c}0\\ \text{-}{\mathrm{C}}_{\mathrm{f}}\mathrm{u}\sqrt{{\mathrm{u}}^2+{\mathrm{v}}^2}\\ \text{-}{\mathrm{C}}_{\mathrm{f}}\mathrm{v}\sqrt{{\mathrm{u}}^2+{\mathrm{v}}^2}\end{array}\right]$$

(3)

In the equations, g represents the gravitational acceleration with a value of 9.81 m/s², h is the water depth, u and v are the depth-averaged velocities in the x- and y-directions, respectively, R is the rainfall intensity, I is the infiltration rate, z is the bed elevation, and C_f is the friction coefficient, determined by ${\mathrm{C}}_{\mathrm{f}}=\mathrm{g}{\mathrm{n}}^2/{\mathrm{h}}^{1/3}$, where n is Manning’s roughness coefficient.

In this study, the widely used modified Horton equation is employed as the method for calculating infiltration.

$${\mathrm{f}}_{\mathrm{p}}={\mathrm{f}}_{\mathrm{c}}+{\left({\mathrm{f}}_0-{\mathrm{f}}_{\mathrm{c}}\right)\mathrm{e}}^{-{\mathrm{k}}_{\mathrm{d}}\mathrm{t}}$$

(4)

In the equation, f_c represents the minimum infiltration rate or steady-state infiltration rate, f₀ is the maximum infiltration rate or initial infiltration rate, and k_d is the decay coefficient. This equation describes the decay process of the infiltration rate over time, reflecting the differences in infiltration capacity between the initial and steady-state phases of rainfall.

The drainage capacity of the urban sewer network is based on the urban drainage planning of each city and is subtracted during the infiltration process. For example, if the drainage standard is designed for a 3-year recurrence interval, the corresponding rainfall intensity is calculated based on the storm event formula, and then the net rainfall input model is used for flood inundation calculations.

The model [34] utilizes structured uniform orthogonal grids for discretization and employs the finite volume method for numerical computation.

2.2.2. The National Model

The EWS consists of two components: the National Urban Rainstorm-Inundation Model and the City-Level Rainstorm-Inundation Model. When the national model detects flood risks in specific cities, it activates the detailed city-level risk warning models for those areas.

In China, among cities tasked with flood prevention, there are 31 key flood control cities and 54 important flood control cities, predominantly located in the southeastern region of the Hu Huanyong Line in central and eastern China. Therefore, the national rainstorm-inundation warning model primarily targets major cities in the central and eastern regions. This includes 231 prefecture-level and above cities across 24 regions, such as Guangdong, Fujian, and Zhejiang provinces.

The elevation data (Figure 3) used in this study are based on the 30 m resolution ASTER GDEM V3 (https://www.gscloud.cn accessed on 20 September 2023), and the land use data (Figure 3) is sourced from the Resource and Environmental Science and Data Center, Chinese Academy of Sciences (https://www.resdc.cn accessed on 20 September 2023).

Based on collected data, including China’s administrative boundary vector maps, terrain elevation data, land use type data, soil type data, and the Outdoor Drainage Design Standard (GB50014-2021) (

Table 1. The design return period of storm sewer.

City Size	Urban Area Type
	Central Urban Area	Non-Central Urban Area	Critical Areas in Central Urban Area	Underground Passages and Sunken Plazas in Central Urban Area
Megacities and Supercities	3~5	2~3	5~10	30~50
Large Cities	2~5	2~3	5~10	20~30
Medium and Small Cities	2~3	2~3	3~5	10~20

), a regular grid with a resolution of 1 km × 1 km was established. Following the research [35,36], infiltration and surface roughness parameters were calibrated using land use type data. Parameter selections are summarized in

Table 2. Manning’s roughness coefficient and infiltration parameters.

Land Use Types	Manning’s n	f0	fc	kd
Cultivated Land	0.05	35	20	3
Forest Land	0.07	25	15	4.8
Grassland	0.045	30	20	3.2
Rivers or Water Bodies	0.025	0	0	0
Urban Construction Land	0.1	2	1	4.8
Unused Land	0.03	40	25	4.8

.

Since this study primarily focuses on urban flooding, particularly the impact of rainfall on cities, the built-up areas were selected as the study sites. The data on urban built-up areas is derived from the China 2020 Built-up Area Dataset published by Sun Jie [37]. A visual representation of built-up areas for select cities is shown in Figure 4.

2.2.3. The Urban Model

Taking Beijing as an example, the construction of the city-level model is explained in detail.

Using Beijing as an example, the construction of a localized urban flood model is explained. Based on the Beijing Urban Master Plan (2016–2035), this study focuses on early warning systems for rainstorm-induced urban flooding in the central districts of Beijing, including Dongcheng, Xicheng, Haidian, Chaoyang, Fengtai, and Shijingshan. To enhance the precision of flood early warnings, the influence of surrounding areas on the central districts was also considered. The modeling area includes the Sixth Ring Road, with an additional expansion to the southwest, as shown in Figure 5a.

A two-dimensional hydrodynamic model was employed to simulate and forecast urban flooding in Beijing’s central districts. The study used a grid resolution of 30 m × 30 m to construct a two-dimensional surface hydrological–hydrodynamic model, generating approximately 4.66 million grid cells. Runoff parameters and surface roughness were assigned based on land use type data. The topography and land use distribution of the study area are illustrated in Figure 5b,c. The elevation data used in this study are based on the 30 m resolution ASTER GDEM V3 (https://www.gscloud.cn), and the land use data is sourced from the Resource and Environmental Science and Data Center, Chinese Academy of Sciences (https://www.resdc.cn).

For major drainage channels in Beijing’s central districts, such as the North Canal, Wenyu River, and Liangshui River, grid elevations were adjusted based on relevant design specifications. Considering the extensive data volume of the urban stormwater drainage network and its impact on computational efficiency, this study utilized a net rainfall input model for inundation calculations. The approach was guided by outcomes from the Beijing Flood and Waterlogging Management Plan and the Flood Control and Waterlogging System Plan for Beijing’s Central Urban Area, incorporating design recurrence intervals of drainage pipelines.

2.3. Early Warning Rules

Based on the severity of urban flooding hazards, flood risk warning levels are classified into four categories: low risk, moderate risk, high risk, and very high risk. These levels are represented by blue, yellow, orange, and red, respectively (e.g., the red warning in risk alerts represents a very high risk). This study integrates early warning criteria from multiple cities, including Beijing, Wuhan, and Shanghai, to define the thresholds for urban flood risk warning levels. The threshold values are summarized in

Table 3. Threshold values for urban flood grid warning.

Serial Number	Warning Level	Urban Flood Risk Warning Threshold for Administrative Districts
1	Blue Alert	5 cm ≤ H, and the area with inundation duration ≥ 1 h accounts for ≥10% of the built-up area.
2	Yellow Alert	(15 cm ≤ H, and inundation duration ≥ 1 h + 5 cm ≤ H, and inundation duration ≥ 3 h) ≥10% of the built-up area.
3	Orange Alert	(30 cm ≤ H, and inundation duration ≥ 1 h + 15 cm ≤ H, and inundation duration ≥ 3 h) ≥10% of the built-up area.
4	Red Alert	(50 cm ≤ H, and inundation duration ≥ 1 h + 30 cm ≤ H, and inundation duration ≥ 3 h) ≥10% of the built-up area.

H is the maximum inundation depth.

.

The overall urban flood warning is based on the grid calculation results from the national model, which includes parameters such as the maximum inundation depth and the longest inundation duration for the city’s built-up area. The area proportion of grids at different warning levels is then calculated to determine the overall urban warning level. For cities classified with yellow or higher warning levels, urban flood warning notifications are issued. Following this, the localized urban model is activated, and depending on the specific needs, the urban flood warning can be further subdivided into district, county, or street-level alerts.

3. Application Effects

3.1. National Early Warnings

Based on the national model, overall early warnings are issued for major cities in the central and eastern regions, and warning levels are classified accordingly. For national urban flood risk early warnings, the rainfall events are considered from 00:00 to 24:00 each day, with flood risk warnings issued for cities in the central and eastern regions of the country on that specific day.

The rainfall data selected includes hourly measured areal rainfall data from five events between 28 July and 1 August 2023. After deducting the city’s drainage capacity, the net rainfall is input into the model for flood simulation. Urban flood warnings are issued based on daily precipitation data for overall city-wide flood risk assessment.

Using the observed rainfall data from 28 July 2023 (Figure 6), flood warning levels for each city are classified based on the inundation depth and duration within the built-up areas. The warning levels are divided into blue, yellow, orange, and red, as shown in Figure 7.

According to reports, Quanzhou experienced flood risks on 28 July 2023. A total of 383 flooded road sections were reported across the city, with an inundated area of 29,180.5 m² in Fengze District. The entire Fuzhou metro network suspended station operations, and some metro lines closed their entrances and exits. Other regions also reported urban flooding situations in the news (Figure 8).

$$\mathrm{ACC}={\displaystyle \frac{\mathrm{p}}{\mathrm{q}}}$$

(5)

In the equation, ACC represents the hit rate; p is the number of cities where the model issued an early warning and flooding actually occurred; and q is the total number of cities for which the model issued a warning.

In this case, flooding situations reported in the news are considered successful warnings. Among the cities with red alerts, Fuzhou, Quanzhou, and Putian had successful warnings, resulting in a hit rate of 60%. All cities with orange alerts had successful warnings, leading to a hit rate of 100%. Among the cities with yellow alerts, nine cities, including Jinan and Xingtai, had successful warnings, giving a hit rate of 62.5%. The overall hit rate for the day was 65.22%.

The forecast results from 29 July to 1 August 2023, are shown in Figure 9. Based on the actual flooding information, the five-day warning results were analyzed. The success rates for the warnings were as follows: red alert at 75%, orange at 80.77%, yellow at 49.45%, and blue at 38.46%. Among these, the success rates for blue and yellow alerts were relatively low. Analysis suggests that this may be due to shallow inundation or the use of emergency measures, such as mobile pump stations, by relevant authorities to quickly drain the water, minimizing the impact on residents’ daily lives and production. As a result, there were no related reports on these instances.

The national urban rainstorm flood risk early warning model has been running at the National Disaster Reduction Center for two years, and the results are considered reliable.

3.2. Urban Early Warnings

After the warning is issued, the urban model is activated. Taking Beijing, a city with a red alert warning, as an example, the urban-level rainstorm flood risk warning is conducted.

3.2.1. Reasonableness Analysis

The precipitation data used for the simulation were collected from over 300 monitoring stations, including Longquan and Dongguoyuan, with the calculation period spanning from 29 July to 2 August.

The simulation results for the maximum inundation depth distribution in central Beijing during the “23·7” rainstorm event are shown in Figure 10. According to the results, 16.96% of the total simulated area had a maximum inundation depth exceeding 0.2 m; 10.3% had a depth exceeding 0.5 m; 4.67% exceeded 1 m; and 1.48% had a depth exceeding 2 m. From Figure 10, it can be observed that the southwestern part of the study area had a relatively large maximum inundation depth, which corresponds to the flood-affected areas of Fangshan District and Mentougou District during the “23·7” Beijing extreme rainstorm event.

Based on the measured heights of the flood traces, the maximum water levels at 88 flooding points within the study area were collected, and the measured water depths were compared with the maximum inundation depths calculated by the model (Figure 11). Among these, 53 inundation points, where the difference between the measured and simulated depths did not exceed 15 cm, accounted for 60.28% of the total points. There were 68 inundation points where the difference did not exceed 30 cm, representing 77.27% of the total points. The predicted inundation points were largely consistent with the actual conditions.

False alarms may have occurred because relevant management departments took timely drainage measures during actual rainfall events, reducing flooding, which the model was unable to capture. On the other hand, missed warnings may be related to insufficient terrain accuracy, especially in low-lying areas prone to flooding, where the model did not fully consider these terrain details.

In addition, the model was validated using typical rainfall events in Zhengzhou, a central Chinese city, and Shenzhen, a southern city. For Zhengzhou, the “7.20” extreme rainstorm was selected, and simulations based on actual rainfall data showed that 68% of inundation points had an error within 15 cm. For Shenzhen, the “9.7” extreme rainstorm was used, with 77% of inundation points having an error within 20 cm. The water levels of all inundation points were obtained through flood trace measurements.

3.2.2. Flood Warning Applications

The study further analyzed the flood risk warning levels within the built-up area of Beijing. The distribution of different warning levels is as follows: red alert grids account for 11.15% of the built-up area, orange and above warning grids make up 15.18%, yellow and above warning grids cover 18.51%, and blue and above warning grids occupy 20.54%. Based on the urban flood risk early warning thresholds, the flood risk warning level for the modeling area in Beijing is classified as a red alert.

Taking the central districts of Beijing as an example, the maximum inundation depth (max_h) is shown in Figure 12, and the proportion of maximum inundation depths and the overall flood risk warning levels are summarized in

Table 4. Proportion of maximum inundation depth and overall flood warning levels.

Area	max_h > 0.2 m	max_h > 0.5 m	max_h > 1 m	max_h > 2 m	Warning Level
Xicheng District	8.89%	2.93%	0.52%	0%	Blue Alert
Dongcheng District	9.36%	3.09%	0.53%	0%	Yellow Alert
Haidian District	16.61%	10.35%	4.55%	1.01%	Red Alert
Chaoyang District	17.59%	9.56%	3.06%	0.46%	Orange Alert
Fengtai District	25.18%	17.67%	8.75%	2.59%	Red Alert
Shijingshan District	20.49%	15.45%	8.96%	2.55%	Red Alert

. Among these, Fengtai District had 33 inundation points, accounting for 37.5% of all inundation points in the simulation area. Of these, 6 points had a water depth exceeding 1 m, and 11 points had a depth exceeding 0.5 m.

For the districts/counties that issued early warnings, street-level flood risk warnings can be further issued. Taking Dongcheng District as an example (Figure 13), the overall flood risk warning was conducted for streets such as Donghuamen Street and Qianmen Street. Due to the high drainage standards in Dongcheng District and the relatively low rainfall intensity, most areas were classified under blue alerts.

4. Discussion and Conclusions

Most urban flood warnings are targeted at small areas or individual cities [38,39], but the models often operate slowly and lack efficiency [40]. Additionally, not all cities have implemented model-based predictions, with only a few having explored this area. There is a pressing need for a model capable of providing rapid flood risk warnings for multiple cities simultaneously. To address this, this study proposes a two-layer EWS model.

The proposed two-layer EWS model consists of national and urban levels. At the national level, it simultaneously computes risk levels for 231 cities and classifies them into warning levels. For cities with warning levels of yellow or above, the urban-level warnings are activated. The urban-level warning system allows for risk warnings tailored to various administrative divisions as needed. In addition, the two-layer EWS model incorporates GPU parallel computing technology, requiring only 5 min for nation-level model computations and just 20 min for urban-level model computations.

However, this study has some limitations. First, infiltration parameters are categorized based on land use, which lacks precision. In areas with complex land use patterns, determining infiltration parameters based on broad land use categories may reduce model accuracy. Further refining land use classifications and incorporating more field-measured infiltration data can help optimize parameter settings and improve accuracy. Second, artificial drainage capacity was not considered. In cases of surface water accumulation, manual interventions often mitigate flood severity. Third, the actual drainage capacity of urban pipelines often deviates from design standards, and these discrepancies were not accounted for when building the model. Additionally, the model generalizes the drainage system, which may reduce its accuracy. Fourth, in complex urban environments, terrain accuracy significantly impacts prediction accuracy. Elevated roads, depressed areas, and other urban features can lead to errors in the prediction results. Finally, the designation of urban warning levels does not consider resilience factors such as flood control infrastructure and emergency drainage measures.

This study proposed a method for urban rainstorm flood risk warning and established a flood risk warning model for key cities in central and eastern China. Through the application of five observed rainfall events from 28 July to 1 August 2023, and the example of the Beijing “31 July 2023” rainstorm-induced urban flooding, the reasonableness of the proposed method was validated. Since the model’s implementation in 2023, it has issued over 100 flood risk warning alerts to the Ministry of Emergency Management each year, providing technical support for urban flood defense at the national level.

However, given the limitations in high-resolution city topography, drainage network data, and other data sources, further research will be needed to refine studies on urban drainage, river boundary conditions, and flood control operations. By incorporating high-resolution DEM data to improve terrain accuracy and combining detailed drainage network information, including various pipe diameters, pumping stations, and gates, the model’s simulation accuracy for drainage capacity can be enhanced.

Future research will involve refining model parameters and conducting a more comprehensive analysis of drainage networks and river characteristics in complex regions to enhance the accuracy and reliability of the urban-level model. Meanwhile, the model will incorporate the simulation of artificial drainage capacity, considering factors such as pipe sedimentation, activation of emergency pumping stations, and the impact of drainage infrastructure aging and maintenance. In addition, future research will focus on analyzing the specific impact of urban floods on vulnerable assets and incorporate factors of urban resilience to optimize urban warning thresholds. This optimization aims to enhance the precision and targeting of risk warnings, thereby improving the practical utility and reliability of the model.

The accurate operation of the EWS relies on high-precision data, including rainfall, terrain, and drainage information, as the quality of the data directly affects the accuracy of the warnings. Therefore, it is essential for cities to provide detailed and reliable data support. Furthermore, the warning thresholds of the EWS should be optimized based on the geographical characteristics, drainage capacity, and historical flood records of each city to ensure the system’s adaptability and effectiveness. Consequently, the implementation of the system requires not only technical support but also collaboration among multiple departments, including meteorology, water management, urban planning, and emergency management. These departments should work together to promote data sharing, threshold setting, and the improvement of emergency response mechanisms, thereby enhancing the city’s flood disaster warning capabilities and disaster prevention level.