Assessing carbon emission impacts of sponge city development: insights from runoff reduction analysis Open Access

The escalating severity of global warming and the increasing frequency of extreme weather events have underscored the urgent need for effective greenhouse gas mitigation and carbon emission reduction strategies. Urban areas, responsible for approximately 80% of global CO₂ emissions (Ghaemi & Smith 2020), play an important role in addressing this challenge. Concurrently, urban deforestation, soil erosion, and wastewater discharges exacerbate greenhouse gas emissions, further emphasizing the crucial importance of exploring green urban development strategies to mitigate carbon emissions. The concept of low impact development (LID), first systematically proposed in the 1990s in Prince George's County, Maryland, USA, is a sustainable urban flood management approach (Gulshad et al. 2024). The approach attempts to minimize the cost of stormwater management by taking a ‘design with nature approach’. LID aims to mitigate the impacts of urbanization and climate change by restoring urban hydrology to pre-development conditions through distributed stormwater controls and natural hydrological elements (Eckart et al. 2017). Building upon this foundation, the concept of sponge cities emerged in China, integrating natural ecosystem principles into urban planning to enhance water cycle management, urban adaptability, and resilience (Zevenbergen et al. 2018). Both sponge cities and LID measures aim to control stormwater runoff and mitigate the adverse effects of urbanization on surface water flow and quality (Liang et al. 2020). Sponge city initiatives represent an expansion and deepening of LID, encompassing a broader range of urban water resource management strategies (Nguyen et al. 2019). The core elements of sponge cities include measures that collect, store, purify, and utilize rainwater; these measures reduce energy consumption in water transport and treatment, achieve runoff and carbon emission reductions, and are widely applied in the design and renovation of urban drainage systems (Qiao et al. 2020). By enhancing surface permeability and the water absorption capacity of soils, technologies such as green roofs, permeable paving, rain gardens, constructed wetlands, and rainwater harvesting and utilization systems increase surface water absorption, reduce runoff, and improve the efficiency of water infiltration and utilization (Han et al. 2023). Lee & Bang (2000) demonstrated how sponge city technology can improve water quality through runoff mitigation; by enhancing the control and utilization of rainwater, it is possible to significantly reduce carbon emissions associated with energy consumption, treatment processes, and water reuse in municipal drainage systems.

Research on carbon emission accounting has predominantly focused on industrial and energy sectors, employing various methods to estimate greenhouse gas emissions, including process boundary examination, baseline emissions, project-based accounting, and models for emission quantification and leakage prediction (Wang et al. 2023a, b). In the context of urban construction and management, carbon emissions research has primarily addressed residential areas and water treatment sectors. For residential sectors, system dynamics models have been utilized to assess the impact of social, economic, energy, and environmental factors on urban carbon emissions and to estimate energy-saving strategies for residential buildings (Xu et al. 2024). Current research on carbon emissions in sponge cities focuses on the emission patterns of individual sponge facilities or city-scale life cycle carbon accounting. Studies utilizing guidelines from the Intergovernmental Panel on Climate Change (IPCC) and life cycle assessment (LCA) have demonstrated that material and energy consumption in sponge city implementation is significantly lower than in integrated urban drainage systems (Zhang et al. 2024). LCA evaluations indicate that rain gardens possess the highest carbon sequestration capacity, effectively offsetting their carbon footprint and significantly reducing carbon emissions (Kavehei et al. 2018). At a macro level, estimating the life cycle carbon emissions of sponge communities, park green spaces, and other urban elements is central to research on carbon emission reduction in sponge cities (Shao et al. 2018). Despite significant progress having been made in carbon emission accounting, research on municipal drainage systems, a key infrastructure, remains relatively underdeveloped. Municipal drainage systems are crucial for urban operations, requiring significant energy for construction and operation, especially for pump stations and power-driven equipment. The construction process involves substantial use of building materials such as cement, steel, and plastic, which generate considerable carbon emissions during their production and transportation (Risch et al. 2015; Lin et al. 2018). Municipal drainage systems also need to handle large volumes of wastewater and stormwater, and their treatment processes can also produce greenhouse gases (Liu et al. 2020). In the water treatment sector, analysis of the carbon footprint of urban wastewater treatment plants (WWTPs) categorizes the process into three stages: pretreatment, biological treatment, and sludge treatment, identifying indirect CO₂ from electricity usage as the major source in the sludge treatment stage (Li et al. 2023). Integrating green infrastructure (GI) such as LID and sponge city initiatives with municipal drainage systems promotes natural infiltration and onsite absorption of stormwater. The integration reduces runoff and wastewater treatment demands, thereby lowering related carbon emissions (Green et al. 2021). The SWMM can simulate sponge city construction, enabling quantification of runoff and carbon reductions (Li et al. 2022). However, there is limited research on the mechanisms by which runoff mitigation leads to carbon emission reduction, and there is a lack of accurate and effective assessment methods to measure the impact of runoff mitigation on carbon emissions. In summary, while significant progress has been made in carbon emission research in the urban construction sector (including municipal drainage systems and sponge cities), such studies have generally been limited to estimating carbon emissions of one or several types of individual facilities or rely on conventional methods such as LCA and carbon emission factors to estimate total regional carbon emissions. These studies lack systematic analysis of the carbon reduction mechanisms of different types of municipal drainage facilities or sponge city facilities at the urban or regional scale. In particular, there is still a gap in research on how sponge city facilities reduce carbon emissions through two primary mechanisms: runoff reduction and rainwater purification. To address this gap, we employed the SWMM model, which effectively simulates regional runoff and pollutant variations, and ingeniously combined the SWMM-based runoff reduction and rainwater purification results with the carbon emission factor method. This approach systematically sorted out the full-process mechanism from runoff reduction caused by sponge city construction to carbon emission reduction and approximated the carbon reduction potential of typical sponge facilities from the perspective of runoff reduction.

Specifically, we assess the carbon reduction efficiency of sponge initiatives before and after implementation, identifying key factors influencing carbon emissions, aiming to further explore the potential of sponge cities for carbon reduction, and improve the efficiency evaluation system of sponge facilities. The methodology, starting from a carbon reduction perspective, calculates the changes in runoff and carbon emissions before and after the implementation of sponge city facilities technologies, providing reliable data support for urban planners and decision-makers. This is crucial for assessing the actual effects of sponge city construction and provides a vital reference for the planning and implementation of similar projects in the future. Consequently, this study fills a gap in research on the mechanisms by which runoff reduction leads to carbon emission reductions, offering an effective method for quantifying carbon reductions to promote sustainable urban development and climate change mitigation. Moreover, this study's combination of quantitative and qualitative methods provides targeted policy recommendations for urban managers, aiming to support scientific decision-making in sponge city planning and construction, optimize resource allocation, and promote a win–win scenario for sustainable urban development and carbon emission control.

Wuhu City, located downstream of the Yangtze River, experiences a subtropical monsoon climate with an average annual precipitation of approximately 1,200 mm. This rainfall is predominantly concentrated between June and September, peaking in mid-June (Table 2). The intensity of precipitation is moderately high nationally, providing a rich water resource base for the application of sponge city technologies. This makes Wuhu suitable for assessing the performance and effectiveness of sponge cities under extreme weather conditions.

The city features a primarily flat area with an average altitude of 10–20 m, interspersed with undulating hills and low mountains, creating a varied terrain that naturally supports the implementation of sponge city features by facilitating effective rainwater infiltration and accumulation. Situated along the Yangtze River, Wuhu has rich water systems, including lakes like Jinghu and Fangtang and wetlands, offering favorable conditions for sponge city functions such as infiltration, detention, storage, purification, utilization, and discharge. Furthermore, the varied soil types, including sandy loam, clay, and silty sand, provide different characteristics for rainwater infiltration and purification, further supporting sponge city technology applications.

To sum up, Wuhu is an ideal location for the construction of a sponge city, likely to achieve satisfactory results. Sponge city construction, through improved urban water management, can reduce urban flooding and water pollution, improve the quality of the urban ecological environment, and enhance the city's sustainable development capacity. Additionally, sponge cities indirectly reduce carbon emissions through runoff reduction and rainwater purification processes, playing a significant role in carbon emission reduction. Currently, few scholars have focused on this aspect, making it necessary to explore the impact of sponge city facilities on carbon reduction from the perspective of runoff reduction.

This study primarily utilized data including pipelines, underlying surface types, topography, and rainfall within the study area, with the specific data and their sources listed in Table 2.

Data processing was divided into three stages: data preprocessing, runoff calculation based on SWMM, and carbon emission accounting (Figure 3). First, pipelines and catchment areas were preprocessed within software (AutoCAD 2014 and ArcGIS 10.2); second, the preprocessed data were imported into SWMM 5.2 for runoff calculation; finally, the carbon emissions were calculated with carbon emission accounting methods.

Given the large volume of computer-aided design (CAD) data and the challenge of directly importing into SWMM, preprocessing was necessary for efficient and more accurate conversion. This involved extracting relevant CAD pipeline data, such as pipeline diameter, length, and invert elevations (both upstream and downstream). To improve model accuracy, the study area was manually segmented into sub-catchment areas based on terrain and underlying surface characteristics. The preprocessed data were then imported into ArcGIS, where nodes were automatically generated, and ground and bottom elevation attributes were assigned to each node. Finally, ArcGIS was used to conduct pipeline attribute extraction, regional flow direction analysis, slope and aspect calculation, and to establish the topological relationships of stormwater flow using spatial union, merging, and association in ArcGIS (Figure 4).

Once ArcGIS data were processed, the final step involved importing the saved personal geodatabase file into SWMM via an access database, thus completing the preliminary model setup.

The SWMM model is adept at calculating the runoff reduction of sponge cities and evaluating the effectiveness of various sponge city facilities in mitigating runoff (Zhang et al. 2020; Li et al. 2022). When combined with the carbon emission factor method, it can approximately estimate the carbon reduction achieved through runoff management and assess the potential carbon reduction of various sponge city facilities (Dong et al. 2023). Before using SWMM to calculate runoff, several assumptions were made to simplify the study, reduce uncertainty, and ensure the reliability of the conclusions.

• (1) Precipitation uniformity assumption: This assumption applies to areas with similar scales, flat terrain, and stable climates. The relatively small spatial extent of the study area (193,718 m²) results in minimal spatial variation in precipitation, which aligns with the uniformity assumption.

• (2) Soil permeability assumption: Based on preliminary soil surveys, it was assumed that the soil in the study area consists entirely of medium-permeability loam or sandy loam, with moderate vegetation coverage and starting infiltration from an unsaturated state.

• (3) Sponge facility efficiency stability assumption: Given the short study period (1 year) without extreme weather or facility damage, this assumption facilitates the analysis of the impact of the facilities on runoff and carbon emissions.

• (4) Surface slope consistency assumption: Considering the overall flat terrain of the study area, surface slope consistency was assumed to simplify flow simulation and avoid excessive complexity. This assumption improves research efficiency without materially affecting results when evaluating the impact of sponge city facilities.

Regarding water quality simulation, the SWMM can simulate the pollutant loads and concentrations, such as suspended solids, nitrogen, phosphorus, and organic matter, covering the entire process of generation, runoff, and migration. The water quality simulation involves the following steps.

First, determine the pollutants to be simulated, such as suspended solids, nitrogen, phosphorus, and organic matter. Second, subdivide the sub-catchment into different land-use areas, as various land-use types generate different pollutant types and growth rates. Third, set pollutant accumulation and scour characteristic functions for each area to simulate pollutant transport in the drainage system. The accumulation process refers to the continuous increase in pollutants caused by natural and human activities before rainfall. SWMM mainly uses power, exponential, and saturation functions for simulation. The scouring process involves soil erosion and partial pollutant dispersion during runoff generation, modeled using exponential functions, performance curves, and event-averaged concentration models (Liu et al. 2024).

The simulations were conducted based on the suspended solids (SS) pollutants recommended by the Sponge City Construction Technical Guide. The pollutant concentrations in natural rainwater were as follows: SS is 10 mg/L, COD is 20 mg/L, TN is 1.0 mg/L, and TP is 0.02 mg/L. Specific water quality simulation parameters are shown in Table 3.

SWMM parameters are divided into physical and empirical parameters. Physical parameters are obtained through measurements and mainly include sub-catchment characteristics (e.g., surface type, width, and impervious area ratio), drainage network features (e.g., size, dimension, and slope), and soil properties. Empirical parameters rely on experimental data and experience, such as storage depth, infiltration parameters, and Manning's roughness coefficient (N-value) (Jiang et al. 2024). Parameters such as pipe length, slope, and diameter were calculated based on the measured terrain data of the Mengxi Ecological Park. The slope was derived from ArcGIS processing of DEM data, with a slope of 1% in the study area. Storage depth reflects the rainwater retention capacity, the N-value represents flow resistance, and the infiltration rate and curve number (CN value) describe soil infiltration and runoff generation characteristics. The initial parameter values were referenced from the literature and the SWMM manual (Rossman 2010) and were determined through calibration. The specific results are shown in Table 4. The calibration steps were as follows: (1) Data input and simulation: Precipitation data from January to March 2024 were input into the SWMM model (Table 2) to simulate runoff under different surface conditions. (2) Measured data collection: The measured data were obtained from ultrasonic flow meters within the study area, providing accurate precipitation monitoring. Additionally, field data on soil and vegetation measurements supported calibration to ensure parameters were both regionally accurate and scientifically valid. (3) Parameter adjustment: By comparing simulated runoff results with the measured data (from monitoring points such as stormwater wells and drainage outlets), parameters like the N-value, infiltration rate, and CN value were adjusted using a manual trial-and-error method. After each adjustment, the model was rerun to observe the deviations. Iteration adjustments were performed until the errors were within an acceptable range, finalizing the calibrated parameters.

Catchment areas were delineated based on surveyed topographic (Figure 1) and underlying surface types (Figure 2), resulting in 493 catchment areas, 141 nodes, and 128 pipes. During the renovation, some hard surfaces were converted to permeable pavements, some regular green spaces, rain gardens, and sunken green spaces, and some hard roofs to green roofs (Figure 2). Plot 1 of the park had 46 pipes, and Plot 2 had 82 pipes. Plots 1 and 2 were two separate areas with no pipe connections between them. The diameters of the stormwater pipes range from 300 to 1,000 mm. The depth of rainwater wells generally ranges from 4 to 6 m. Overflow storage ponds were constructed on the west side of the park, with outlets connected to the wastewater treatment plant via pump stations. The results of the model creation are shown in Figure 4(e). After the model was constructed, surveyed precipitation data were imported into the model to calculate runoff, the results of which are presented in Table 6.

The carbon emission accounting process primarily involves defining the accounting subjects, selecting methodologies, obtaining activity data, and choosing emission factors (Shan et al. 2020). This includes energy consumption at pump stations (Singh & Kansal 2018), wastewater treatment processes (Zib et al. 2021), and the process of overflow from combined sewer pipelines into water bodies (Rizzo et al. 2020).

• (1) The carbon emissions from the power consumption of the pumping station refer to the carbon emissions generated when the pumping station pressurizes and transports the initial rainwater runoff to the wastewater treatment plant. The calculation method is shown in Formula (5).

• (2) The carbon emissions from the wastewater treatment process refer to the carbon emissions produced by the wastewater treatment plant when processing runoff rainwater, primarily including CO₂, CH₄, and N₂O. The carbon emissions were calculated using Formulas (6)–(9).

• (3) The process of combined sewer overflow (CSO) into water bodies refers to the phenomenon during heavy precipitation when the urban drainage exceeds the capacity of combined sewers or WWTPs, causing untreated rainwater and wastewater to be directly discharged into nearby natural water bodies. Since the overflow wastewater contains untreated organic matter and other substances, greenhouse gases, mainly CH₄ and N₂O, were generated by microbial action in the natural water bodies. The carbon emissions were calculated using Formulas (10)–(12).

According to the measurements, before the renovation of the park, when the precipitation peaked and formed runoff, the flow into the wastewater treatment plant was about three times the flow of the overflowing wastewater into the natural water bodies. After the renovation, the overflowing rainwater was stored in the overflow storage ponds, and this process did not generate carbon emissions. The primary accounting method employed the carbon emission factor method, with specific calculation formulas provided in Table 5.

Table 5

Methods and formulas for carbon emission accounting in the drainage system and key parameters

Category .	Formula .	Note .
Pumping station	(5)	represents the carbon emission during the power consumption process of the pumping station (kg CO2 eq); take 1,000 kg/m3 as the density of water (kg/m3); g is the gravitational acceleration, 9.8 m/s2; is the elevation difference (m) between the start and end points of water transportation; is the head loss along the pipelines network (m), where v, d, and L are the flow velocity, diameter, and length of the water supply pipelines, in units of m/s, m, and m, respectively; is the initial precipitation amount; is the efficiency of water pump operation; is the electricity emission factor for the region (kg CO2 eq/kWh).
Wastewater treatment plant	(6) (7) (8) (9)	is the carbon emission of CO2 in the operating process of the wastewater treatment plant (kg CO2 eq); is the carbon emission of CH4 in .the operating process of the wastewater treatment plant (kg CO2 eq); is the carbon emission of N2O in the operating process of the wastewater treatment plant (kg CO2 eq); is the amount of COD contained in rainwater treated by the wastewater treatment plant (kg/m3); is the amount of TN contained in rainwater treated by the wastewater treatment plant (kg/m3); is the initial precipitation amount treated by the waste treatment plant (m3); is the carbon emission factor (kg CH4/kg COD) of CO2 gas in the wastewater treatment process; is the carbon emission factor (kg CH4/kg COD)of CH4 gas in the wastewater treatment process; 28 is the global warming potential energy value of CH4, constant (kg CO2 eq/kg CH4); is the carbon emission factor (kg CH4/kg TN) of N2O gas in the wastewater treatment process; 265 is the global warming potential energy value of N2O, constant (kg CO2 eq/kg N2O); 22/14 is the molecular weight ratio of 1/2 N2O to N; E2 is the carbon emission during the wastewater treatment process (kg CO2 eq)
CSO	(10) (11) (12)	is the carbon emission from CH4 in wastewater overflow (kg CO2 eq); is the COD concentration in overflow wastewater (kg COD/m3); is the total amount of overflow wastewater; is the CH4 emission factor after wastewater overflow into water bodies (kg CH4/kg COD); 28 is the global warming potential of CH4, a constant (kg CO2 eq/kg CH4); is the carbon emission from N2O in wastewater overflow (kg CO2 eq); is the TN concentration in overflow wastewater (kg N/m3); is the N2O emission factor after wastewater overflow into water bodies (kg N2O/kg COD); 265 is the global warming potential of N2O; 22/14 is the molecular weight ratio of 1/2 N2O to N; E3 is the carbon reduction during the regulation process (kg CO2 eq).

Category .	Formula .	Note .
Pumping station	(5)	represents the carbon emission during the power consumption process of the pumping station (kg CO2 eq); take 1,000 kg/m3 as the density of water (kg/m3); g is the gravitational acceleration, 9.8 m/s2; is the elevation difference (m) between the start and end points of water transportation; is the head loss along the pipelines network (m), where v, d, and L are the flow velocity, diameter, and length of the water supply pipelines, in units of m/s, m, and m, respectively; is the initial precipitation amount; is the efficiency of water pump operation; is the electricity emission factor for the region (kg CO2 eq/kWh).
Wastewater treatment plant	(6) (7) (8) (9)	is the carbon emission of CO2 in the operating process of the wastewater treatment plant (kg CO2 eq); is the carbon emission of CH4 in .the operating process of the wastewater treatment plant (kg CO2 eq); is the carbon emission of N2O in the operating process of the wastewater treatment plant (kg CO2 eq); is the amount of COD contained in rainwater treated by the wastewater treatment plant (kg/m3); is the amount of TN contained in rainwater treated by the wastewater treatment plant (kg/m3); is the initial precipitation amount treated by the waste treatment plant (m3); is the carbon emission factor (kg CH4/kg COD) of CO2 gas in the wastewater treatment process; is the carbon emission factor (kg CH4/kg COD)of CH4 gas in the wastewater treatment process; 28 is the global warming potential energy value of CH4, constant (kg CO2 eq/kg CH4); is the carbon emission factor (kg CH4/kg TN) of N2O gas in the wastewater treatment process; 265 is the global warming potential energy value of N2O, constant (kg CO2 eq/kg N2O); 22/14 is the molecular weight ratio of 1/2 N2O to N; E2 is the carbon emission during the wastewater treatment process (kg CO2 eq)
CSO	(10) (11) (12)	is the carbon emission from CH4 in wastewater overflow (kg CO2 eq); is the COD concentration in overflow wastewater (kg COD/m3); is the total amount of overflow wastewater; is the CH4 emission factor after wastewater overflow into water bodies (kg CH4/kg COD); 28 is the global warming potential of CH4, a constant (kg CO2 eq/kg CH4); is the carbon emission from N2O in wastewater overflow (kg CO2 eq); is the TN concentration in overflow wastewater (kg N/m3); is the N2O emission factor after wastewater overflow into water bodies (kg N2O/kg COD); 265 is the global warming potential of N2O; 22/14 is the molecular weight ratio of 1/2 N2O to N; E3 is the carbon reduction during the regulation process (kg CO2 eq).

View Large

is the peak runoff (m³/s); C is the runoff coefficient of different underlying surface types, depending on the soil properties of the watershed; I is the intensity of precipitation (mm/h); and A is the area of the watershed (m²).

ArcGIS was used to measure the land cover of different underlying surfaces in the park. Given the park's high soil moisture and strong water retention properties, appropriate runoff coefficients were selected. Specific runoff coefficients and areas are listed in Table 6.

Overall, the SWMM model's simulation results closely align with those from the Rational method, with a correlation of 0.9984. When precipitation was low, the runoff simulated by the SWMM model was generally consistent with that calculated using the Rational method; however, as precipitation increases, the runoff simulated by the SWMM model tends to be slightly lower than that calculated by the formula. First, this may be due to discrepancies in the runoff coefficients for different underlying surfaces; second, the Rational method is based on a simpler principle, whereas the SWMM model simulates more realistic and complex precipitation scenarios. In comparison, the SWMM model simulates precipitation more accurately and realistically. Therefore, the simulation results of the SWMM model were considered reliable. And given that the elevation within Mengxi Ecological Park is approximately 7.00 m, and the surrounding elevation is between 6.20 and 6.75 m, Mengxi Ecological Park is higher than its surroundings (Figure 1). Consequently, the park has become largely insulated from external runoff following the implementation of sponge city facilities. Therefore, calculating carbon emission reductions by simulating surface runoff reductions in Mengxi Ecological Park was reliable.

Table 7 summarizes the simulation results before and after the renovation. After the renovation, the runoff volume decreased by 100,840 m³, with reductions contributed by various sponge facilities as follows: green roof (26,966.09 m³), permeable pavement (16,255.00 m³), sunken green space (12,240.54 m³), rain garden (9,499.37 m³), and overflow retention pond (35,879.00 m³). The reductions in COD and TN concentrations were 3,940.69 and 137.04 kg, respectively. Specifically, the COD reductions included 2,252.52 kg from inflows to the sewage treatment plant and 1,688.17 kg from CSO. The TN included 76.46 and 60.58 kg from the same categories. Based on these reductions in runoff and pollutant loads, the carbon reduction of the sponge facilities was calculated. Further analysis shows that, after the modification, the runoff per square meter of land decreased by approximately 52.1%, demonstrating the high efficiency of sponge city designs in rainwater management (Hou et al. 2020). The reductions in COD and TN were 0.0203 and 0.00071 kg, respectively, highlighting the effectiveness of sponge city facilities in reducing pollutant and nitrogen losses. Compared to pre-transformation, the reduced runoff accounted for 82.72% of the total runoff, reflecting the significant improvement in rainwater retention and infiltration capacity, alleviating pressure on urban drainage systems, increasing water resource utilization, and reducing flood risk. The reductions in COD and TN accounted for 69.79 and 56.55% of the total pollutants before the renovation, respectively, indicating the significant role of sponge city facilities in improving water quality (Xiong et al. 2021).

The runoff calculation results show that the sponge city facilities significantly improved the surface runoff conditions in the study area. These findings offer a reference for water resource management and urban planning in cities with similar geographical and climatic conditions, providing data support to enhance urban ecological resilience.

It can be observed that the carbon emission reductions from the CSO treatment were the largest, with a reduction of 3,340.42 kg CO₂ eq (Figure 6). The smallest carbon emission reductions were from the electricity consumption of the pump station, totaling 1,311.43 kg CO₂ eq. It is evident that the park lacked effective stormwater control measures before renovation and utilized combined sewer systems, leading to substantial discharge of both rainwater and wastewater into natural water bodies, thus releasing significant amounts of greenhouse gases. Constructing overflow storage ponds within the park could effectively mitigate this issue.

The SWMM model is widely used to simulate the hydrological cycles of flow and water quality of urban watersheds, including flow and water quality, and is applicable to both single and continuous rainfall events, as well as combined and separated drainage systems. The model can analyze the runoff and treatment processes of various pollutants and simulate the hydrological characteristics of infiltration, storage, and LID facilities (Lee et al. 2022). The LID module supports applications across various land-use types and allows for the setup of scenarios based on infiltration, water storage, or a combination of both. It calculates the inflow and outflow volumes for sub-regions based on the principles of water balance and analyzes the regulatory effects of LID facilities under different rainfall conditions (Bai et al. 2018).

Despite the important role of the SWMM model in runoff and carbon emission analysis, there are still some limitations and uncertainties. First, the model requires high-quality data, and in practice, it is difficult to obtain accurate data on terrain, land use, pipeline networks, and hydrological parameters, especially in older urban areas. Data gaps or inaccuracies can affect model precision (Lee et al. 2022). Additionally, carbon emission factors are influenced by factors such as wastewater treatment scale, inflow quality, operational efficiency, and regional differences. Second, model calibration and validation are complex, and precise calibration data is difficult to obtain. Defining model performance functions is another challenge (Sytsma et al. 2022). Finally, simplifications in the model may affect the accuracy of the results. The actual hydrological cycle and pollution processes are dynamically influenced by multiple factors, and simplified assumptions such as uniform precipitation, soil permeability characteristics, and facility efficiency may fail under certain conditions, leading to evaluation biases. Future research should focus on the integration of multi-source data and the application of intelligent algorithms, using satellite remote sensing, soil moisture data, and machine learning to optimize parameters. This will expand the applicability of the model, improve the precision of sponge city carbon emission studies, and provide stronger support for urban sustainable planning.

This study used the SWMM model in conjunction with the IPCC greenhouse gas emission factor accounting method to analyze the impact of sponge city construction on carbon emissions from the perspective of runoff reduction. We developed a novel carbon emission accounting method to assess the carbon reduction potential of sponge cities. Using Wuhu City's Mengxi Ecological Park as a case study, we conducted a comprehensive analysis of runoff reduction and carbon emissions over a 1-year period, comparing before and after the sponge facility renovation, as well as the runoff reduction, carbon reduction, and carbon reduction efficiency of different sponge city facilities, based on field surveys and the SWMM model. Key conclusions from this study can be summarized as follows:

• (1) We have developed a robust method for calculating the carbon reduction from runoff reduction in sponge cities. This method leverages the SWMM's capability to simulate complex municipal drainage systems, allowing for accurate quantification of the runoff reduction in both traditional and sponge measure systems during precipitation events. By extension, this enables the calculation of associated carbon emissions. Our method provides a valuable technical supplement for quantifying and evaluating the carbon emissions in urban drainage systems, addressing a critical gap in current assessment tools.

• (2) Our SWMM-based estimations revealed significant variations in the runoff mitigation capacities of different sponge city facilities. Specifically, the overflow storage ponds had the highest impact on runoff mitigation, followed by rain gardens, while sunken green spaces had the least impact. These findings underscore the importance of prioritizing the development of overflow storage ponds and rain gardens in sponge city construction to more effectively reduce runoff and mitigate flood risks.

• (3) Using the SWMM model and the carbon emission factor method, we calculated a total carbon emission reduced of 7089.85 kg CO₂ eq within 1-year post-renovation. Among the various measures, the overflow storage ponds and rain gardens showed higher potential for carbon reduction, while sunken green spaces exhibited relatively weaker performance in terms of carbon reduction.

• (1) Emphasize carbon emission reduction in sponge city construction. In sponge city construction, attention should be given to controlling and reducing carbon emissions. Planners need to assess the current carbon emission status and influencing factors in each region and optimize facility layout based on terrain, climate, and land use. For example, in areas with concentrated rainfall and flat terrain, rain gardens and overflow retention ponds should be prioritized to maximize their role in reducing runoff and carbon emissions.

• (2) Quantify carbon emission reduction evaluation for sponge city construction. Use the SWMM model to establish a monitoring system and optimize the calculation process for carbon reduction in sponge city construction, providing a comprehensive indicator system for evaluating the effectiveness of sponge city construction.

• (3) Balance the economic and environmental impacts. Considering the differences in economic and environmental benefits of various types of sponge facilities, such as green roofs, which offer higher carbon reduction benefits but come with higher economic costs, while rain gardens perform excellently in water quality purification and carbon reduction but may lead to mosquito breeding and secondary pollution if not properly designed. These conflicts need to be further addressed in the design and construction of sponge cities at the urban scale in the future.

This work is supported by the National Natural Science Foundation of China (41901129) and the University Natural Sciences Research Project of the Anhui Educational Committee (KJ2024JD22). Zheng Duan acknowledges the support from the Joint China-Sweden Mobility Grant funded by NSFC and STINT (CH2019-8250).

All authors contributed to the study's conception and design. Material preparation, data collection, and analysis were performed by W.L., J.C., H.L., and Z.D. The first draft of the manuscript was written by W.L., J.C., and H.L., and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.