Research on the Dynamic Evolution of Sponge City Construction Risk Based on System Dynamics

Following the research idea of "theoretical analysis -- mechanism analysis -- model simulation", the basic concept and basic theory of sponge city construction risk are rstly dened, and then the risk factors and internal mechanism of sponge city construction are analyzed by the grounded theory method. Finally, the system dynamics model is used for simulation. The dynamic development and key risk factors of each sub-risk system in the whole life cycle are analyzed, and the countermeasures to reduce the risk of sponge city construction are given.


Introduction
Climate observation data in the past 100 years show that the global climate is warming year by year, and the frequency of extreme rainfall is expected to continue to increase in the future (Shastri et al. 2019), leading to huge oods (Zolch et al. 2017). With the acceleration of urbanization, the increase of population density, building density and industrialization degree, urban water consumption and water frequency increase, water resources sharply reduced; The substitution of impervious pavement for natural ground such as vegetation and soil leads to the destruction of urban water ecosystem (Stovin et  serious water ecological problems such as water logging, water system pollution and habitat loss of aquatic organisms (Technical Guide for Sponge City Construction 2014). The Ministry of Construction surveyed 351 Chinese cities in 2010 and found that more than 60 percent had experienced water logging in the previous year, and nearly 40 percent had experienced more than three water logging disasters. An evaluation of more than 700 rivers by the Ministry of Water Resources found that 46 percent of them were polluted, and the country's seven major river basins are facing serious problems such as water shortage and water pollution. Water resources, water security and water ecology interwoven situation (Seyyed and Zahra 2012;Gerald 2003). With reference to the urban storm water management concept of developed countries, China has put forward a unique water treatment scheme --sponge city. Up to now, although the 30 pilot sponge cities in China have achieved obvious results, there is still a certain gap from the expected results. A series of problems, such as imperfect legal norms, imperfect management system and lack of technical experience, have been exposed in the construction process, resulting in the gradual decline of some sponge facilities, poor water logging control effect and the decline of the enthusiasm of the masses. Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js Based on this background, this research to sponge city construction in jiangxi province as the research object, follow the "theoretical analysis, mechanism analysis, and model simulation research train of thought, rst of all, introduce the basic concept of sponge urban construction risk and the basic theory, then through the method of grounded theory to explore risk factors for urban construction, sponge using system dynamics of relevant knowledge (Yizhuang 2007;Shuhua 2002; Kangyi et al. 1987; Ve e 1996), build dynamic evolution model of the urban construction risk, sponge using Vensim PLE software carries on the simulation experiments, through sensitivity analysis determine the main risk factors in each stage of whole life cycle and the in uence mechanism between each system, and the risk, to simulate the results of different countermeasures and discusses the countermeasures should be taken to reduce the sponge urban construction risk, in order to put forward some suggestions for sponge city construction in the future. The overall research framework is shown in (2016) pointed out that the difference between sponge city and traditional urban construction mode is that the latter brings destructive construction to the city, while the former emphasizes that the impact on the environment should be reduced during urban development.
However, some scholars point out that low-impact development is only a small part of sponge city construction (Danjie et al. 2016; Kongjian 2016), which is not equal to "big sponge" facilities such as simple drainage and water logging prevention, nor "small sponge" facilities such as rainwater reuse Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js identi es and evaluates the risks applied in LID project construction with a variety of analytical methods, prioritises the relatively important risks with analytic hierarchy process, and nally puts forward suggestions on risk treatment. Xiang Pengcheng et al. (2018) conducted risk assessment on sponge city construction of Yuelai new city in Chongqing by combining intuitional fuzzy analytic hierarchy Process and grey theory, and believed that economic risk was the most in uential factor, followed by management factors.

Research status of dynamic evolution of project risk
Western scholars are the rst to realize that the traditional static risk management system is not suitable for risk prevention and control of long-span projects. Tummala (1994) pointed out that the risk management process is a dynamic cyclic process and advocated continuous risk management at different stages of the whole project cycle. Jaafari (2000;2001) also proposed that risk management at different stages of a project should serve the goal of the whole project life cycle. Feng Yahong (2008), Zhai Yongjun and Lu Huimin (2007) believe that modern management industry urgently requires comprehensive identi cation and evaluation of project life-cycle risks in a more systematic and dynamic way. In terms of speci c application, Ning Liang and Zhao Libo (2016) constructed an evaluation index system for risk factors of public service outsourcing, and made quantitative evaluation of the risks of the government's implementation of public service outsourcing by using fuzzy comprehensive evaluation method, and analyzed various risk factors affecting the success of outsourcing. Qiu Guangzhen and Xu Shulong (1998) took time as the variable and created six basic functions to describe different risk trends, and used analytic hierarchy process (AHP) to calculate the ranking weight of risk in uencing factors of engineering projects. Sun Chengshuang and Wang Yaowu (2003) proposed the identi cation and monitoring method of dynamic risk factors of construction projects based on the time sensitivity of risk factors and the project income as the evaluation standard. Zhu Kun, Yang Jiaben and Chai Yueting (2004) put forward the concept of "risk bubble" and studied the dynamic transformation mechanism of risk factors theoretically by using the formalized language of risk bubble energy. Liu Wu, Du Zhida and Zhang Qiuyue (2008) took into account the in uence of temporal and spatial changes of risks on construction, calculated network time parameters and Monte Carlo Simulation(MCS) on the basis of PERT network model, and obtained the in uence index curve of each factor according to the principle of project impact.Thus, the dynamic risk analysis of project schedule is carried out.

Review of research status
Sponge city construction projects are time-consuming and involve many stakeholders. According to the current research situation, it is urgent to conduct in-depth research. Citespace software was used to analyze the knowledge graph of literatures related to sponge city construction risk, and collate the publication trend of literatures related to sponge city construction risk in the past decade. The results are shown in Figure 2. Through combing relevant literatures and theories on in uencing factors of sponge city construction and dynamic evolution of engineering project risks, it is found that :(1) sponge city risk research needs to be strengthened in the application of risk identi cation theory and system dynamics Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js theory and other related basic theories. There are many researches on risk decision-making and management of single urban construction project in China, and they focus on the single dimension of economy and society. However, few studies have been done on the identi cation of the complex background and risks peculiar to such group construction projects as sponge city. (2) When studying the risks related to sponge cities and construction, domestic and foreign scholars are mostly problemoriented, studying the causes and manifestations of problems in the construction process and proposing solutions. However, they do not pay enough attention to the multi-stage development characteristics of storm water management risks. (3) The risk of sponge city construction is a dynamic system with variability, and the existing studies have not formulated corresponding countermeasures according to the different characteristics of each stage. In addition, in the traditional security system, they are all suggestions, without empirical consideration of the implementation effect of the scheme.

Sponge City
Sponge city is in response to climate change and rapid urbanization on the city caused by the negative in uence of major urban construction strategy, as well as to realize the sustainable development of urban environmental construction of key concept, is refers to the use of the natural process to manage storm water runoff, which make the city to adapt to the environment change and respond to natural disasters, etc, like a sponge, it has good "elasticity" (Vogel et al. 2015). The action mechanism of sponge city is shown in Fig. 3.

Risk of sponge city construction
Sponge city construction risk refers to the factors that fail to achieve the expected goals due to adverse results caused by various elements and their combination and interaction within the whole life cycle of sponge city project initiation, planning and design, construction, completion acceptance and later operation and maintenance. The probability of risk occurrence and the severity of risk loss are regarded as the risk evaluation criteria.

system theory
The basic idea of system theory was rst proposed by the biologist Bertalanfy in the early 1920s. In the later development process, it was supplemented by the mutation theory of The French mathematician Thom and the synergy theory of the German physicist Haken, and gradually formed a set of theoretical methods with its own characteristics. And it has been applied to the research of various disciplines (Qianhu et al. 2018). Bertalanfy (2007) believes that a system is a collection of interconnected and interacting elements, rather than a simple addition of various elements. Qian Xuesen (2002), a Chinese scholar, further supplemented and deepened the de nition of system, believing that it is an organic whole with speci c functions synthesized by several components that interact and depend on each other, and a part of a larger system to which it belongs.
Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js System theory emphasizes the role of elements, levels, structures, functions and other factors in the system as well as the mutual relationship between various parts (Mees et al. 2013), and calls such relationship "coupling" (Yongwei et al. 2018).

Full life cycle
Every engineering project has speci c development stages, and the speci c de nition and division of each stage changes with different projects, and these stages constitute the whole life cycle of engineering projects . According to the process that a project goes through from start to nish, the whole life cycle theory divides a project into ve stages: "identifying requirements, proposing solutions, implementing the project, accepting the project and closing the project". Based on the reading of relevant literature, this study divides the whole life cycle of sponge city construction into ve stages according to the principles of consistency in time, similarity in content and integrity of project results, as follows: Project start-up stage (pss), planning and design stage (pds), construction and construction stage (ccs), completion and acceptance stage (cas) and operation and maintenance stage (oms).

Risk Identi cation Of Sponge City Construction Based On Grounded Theory
The American Project Management Institute divides the process of risk management into the following parts: risk identi cation, qualitative and quantitative analysis, strategy development and risk monitoring. Among them, risk identi cation is the basis of risk dynamic evolution mechanism research and refers to the process of risk judgment, classi cation and identi cation.

Selection of risk identi cation methods
At present, the risk identi cation methods commonly used in the academic circle can be roughly divided into two categories: analysis method and expert investigation method. Analysis methods include fault tree analysis, WBS work decomposition method, check list method and case analysis method, etc. Expert investigation methods include Delphi method, brainstorming method and scene analysis method. Different risk identi cation methods have different advantages and disadvantages (see Appendix 1) and are also suitable for different scenarios. Considering the characteristics of sponge city construction risks, this paper chooses grounded theory as the risk identi cation method, and the speci c application steps are shown in Fig. 4 according to the topic, citation volume, download volume and time. Related literatures were retrieved from the core library Of Web Of Science, and 15 English literatures were selected. Interviews were conducted with relevant personnel in sponge city pilot cities to further supplement risk factors in sponge city construction.

Risk identi cation (1) Initial coding
Coding is the basic step of grounded theory to analyze qualitative data (Xiaoe 2011). To adopt the way of every code in Chinese and English literature and interview the statement in conceptual analysis, some of the more obscure and inconsistent after interview records deleted, try to use the original literature and the participant's words as code name to minimize the effects of the researchers' subjective bias, the resulting initial code 142. See Appendix 2 for the initial coding and editing process of literature and interview data.
(2) Focus coding At this stage, according to the frequency of the initial coding in the original data mentioned above, after continuous comparison and correction with the original data, the codes with similar meanings are summarized and integrated, and a total of 23 focused codes are obtained. After a preliminary conceptualization of the initial coding, ve main categories including regulations, management, capital, technology and society are summarized and generated. See Appendix 3 for the coding integration process.

Identi cation results
After the above coding process, use the reserved literature theoretical saturation test and interview, showed the study theory of generic has been rich enough, the sponge effect of ve main categories of the construction of the city did not form a new important category, and internal also did not form a new form factor, combined with expert for determine risk factors of life cycle, The nal risk identi cation results are shown in Table 1 After the logical relationship between major risks and boundary factors is fully grasped in the theoretical coding stage, all risk factors are integrated into a system, and the causal loop diagram is drawn as shown in Fig. 6. The arrow in Fig. 6 indicates that there is a causal relationship between the two variables, while "+" and "-" respectively indicate that there is a positive causal relationship and a negative causal relationship between the two factors.

Stock ow chart
The difference between the stock ow diagram and the causal loop diagram is that the causal loop diagram is a static diagram describing the logical relationship of risks, while the numerical and mathematical formulas added to the stock ow diagram can express the dynamic changes of risks in the construction process. To summarize the interaction relationship between risk factors, and according to the above drawing of the causal loop diagram, will "sponge urban construction risk" as the state variables of the total system, construct a state variable, rate and auxiliary variables of three variables in the form of urban construction risk system dynamics sponge stock ow diagram, as shown in Fig. 7 However, G1 method can fully re ect the subjective consciousness of the evaluator, take the subjective will of the decision maker into consideration, and nd out the internal connection of the importance of risk factors on the basis of the sequential relationship between risk factors without testing consistency (Xuejun and Yajun 2006). The risks of Sponge city PPP projects are complex and di cult to be quanti ed. The pure subjective or objective weighting will have an impact on the subsequent risk evaluation, while the integrated weighting method integrates the advantages of both subjective and objective weighting methods, which can not only re ect the subjective will of the chooser, but also apply objective theories and methods. The steps to determine the weight of the integrated weighting method are as follows: W j represents the weight of the risk factor j after the comprehensive integration of the two weighting methods, where W j 1 represents the weight calculated by entropy method and W j 2 represents the weight calculated by G1 method, namely: Where, α is the proportion of objective weight to combination weight. Taking the objective of minimizing the sum of squares of weight deviations by subjective and objective methods, the objective function is established as follows: It can be seen that the integrated weight result is optimal when the subjective weight and the objective weight account for half respectively. By inputting the numerical and dynamic equations of risk factors obtained in the above calculation process into Vensim PLE software, the system simulation diagram of 23 risk factors and 5 risk subsystems changing over time in the whole life cycle can be obtained, and the in uence of each key risk factor on the risk subsystem can be identi ed through sensitivity analysis. However, in order to intuitively identify the risk subsystems that each stage has a great impact on the risk of sponge city construction, the risk level of each stage is summarized by using radar chart.
Radar chart, also known as spider web chart, is often used in the analysis of nancial statements, with intuitive and clear advantages. This method is used for reference in the risk analysis of sponge city construction. Appropriate coordinate axes are established to represent the simulation results of each key risk factor in ve stages of the whole life cycle, and the main risk factors in a certain stage are obtained.
5 Empirical Study --Taking Sponge City Construction In Jiangxi Province As An Example

Overview of the study area
Pingxiang, Jiangxi Province, China, is located in the Hunan-Jiangxi watershed, which is a typical hilly region in the South of the Yangtze River. Most of the basin is mountainous and hilly, and only 11% is plain valley. During the rainstorm, the mountain ood comes rapidly, the river level rises sharply, and the phenomenon of river embankment over ow occurs in the plain valley. After the rain, the mountain ood subsided, the river lack of water supply, river ow is small, dry season almost dry. There are 4 historical water logging areas in the old city, and 84 low-lying potential water logging points are at high risk of water logging. Take wanlongwan water logging area with the most severe water logging as an example. On July 8, 2016, it rained 79.8 mm. Wanlongwan waterlogged an area of 1.2 square kilometers, with the maximum water depth exceeding 1 meter. The contradiction between city and nature, human and water is prominent. In view of the reality of frequent ood disasters and water shortage coexisting due to the hydro logical characteristics of the typical jiangnan hilly region, Pingxiang creatively proposed the systematic construction idea of "whole-domain control --system construction --zoning management".
Since the pilot project, sponge city construction in Jiangxi province has made some achievements, but the effect is far from ideal. Case based on this, through eld visits and consulting for practitioners, found in the process of the sponge construction of city in jiangxi province, the imperfection of the laws, regulations and policies concerning the mechanism of management to organize the implementation of e cient enough, related technology is not mature, inadequate social participation degree, are the construction and maintenance of sponge carries risks.

Scoring boundary risk factors
After tting the risk ow diagram of sponge city construction project, the boundary risk factors were identi ed, and the assignment values of 23 key risk factors and 5 risk subsystems were obtained through structural interview. The questionnaire was distributed to relevant practitioners and university researchers Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js in the pilot cities of sponge cities. A total of 128 questionnaires were distributed, and 93 effective questionnaires were recovered, with an effective recovery rate of 73%. The reliability of the questionnaire was analyzed by SPSS software. The Cronbach coe cient (Cronbach) was 0.873, which had a high reliability. KMO = 0.79 > 0.7, signi cance level P = 0.000 < 0.01, indicating that the data is relatively valid and can be processed in the next step.
According to the above discussion on the theoretical basis, the possibility of risk occurrence and the severity of consequences are mainly considered, and the probability of risk occurrence and severity of consequences are divided into ve grades, with 1-5 points respectively, with 1 point indicating the lowest degree and 5 points indicating the highest degree.The risk factor function is R = f(p, r). According to the multi-factor multiplication comprehensive analysis model commonly used in academia, its calculation formula is as follows: Where, R ij represents the size score of the subject j on the risk factor i; P ij represents the score result of the respondent j on the occurrence possibility of the risk factor i; C ij represents the score of the respondent j on the risk degree of the risk factor consequence i. The initial value of this risk factor was obtained by averaging the scores of all survey experts, and the calculation formula was as follows: Where, n represents the number of respondents; R ij represents the scoring result of the respondent j on the risk of the risk factor i; R represents the initial value of the state variable.
The initial values of the risk factors obtained after averaging the values assigned to each risk factor by each expert (see Appendix 4 for details) are shown in Table 2. preference coe cient is 0.5, the sum of squares of the deviation between the nal weight and the subjective and objective weight is the smallest. Therefore, the average value of the objective weight and subjective weight is taken as the nal weight result, and the speci c result is shown in Table 3.

Establish the equations of each subsystem
The weight assignment obtained from Table 3

Analysis of empirical results
After consulting relevant practitioners of sponge city construction in Jiangxi Province, it is found that the start-up phase is 1 month, the planning and design phase is 2 months, the construction phase is 6 months, and the operation and maintenance warranty period is 2 years. In order to have a better simulation effect, the simulation duration of the model is set as 15 months to simulate the risk changes from project approval to six months of operation.

Full life cycle simulation analysis of regulatory risks
In regulatory risk in this subsystem, incomplete "construction technology standard" and "operation maintenance rules are not perfect" both of these factors will exist in the construction and operational stage, so the time function as constraints, namely in the planning and design phase of the risk factors of risk value is set to 0, do so with the rest of the time the functions below. In order to further explore the dynamic impact of each risk factor on the regulatory risk subsystem, the initial value of each factor was reduced by 30% and other factors remained unchanged for sensitivity analysis. The adjustment scheme is shown in Table 4, and the nal analysis result is shown in Fig. 9 (a).  Keep the initial values of each factor unchanged A2 Initial var of a 5 is reduced by 30% A3 Initial var of a 1 is reduced by 30% A4 Initial var of a 2 is reduced by 30% A5 Initial var of a 4 is reduced by 30% A6 Initial var of a 3 is reduced by 30% Through sensitivity analysis, it can be found that regulatory risk is most sensitive to a 1 factor, followed by a 5 , and least sensitive to a 4 , indicating that promoting the construction of national regulations and the improvement of investment and nancing laws in the start-up stage has a signi cant impact on the reduction of regulatory risk level.
The low sensitivity of factor a 4 is mainly due to the fact that this factor begins to affect sponge city construction in the last stage of the whole life cycle. However, it can be seen from the change trend of each scheme after 9 months that a 4 factor has a extreme in uence on the decrease of regulatory risk level. In order to better analyze this factor, the model of the simulated time extended to 30 months, found that the factors effect on the risk regulations is not lower than the other factors except a 1 , this shows that in the future urban construction run longer sponge maintenance phase, a 4 will risk occupy larger in uence on construction of city sponge weights, as shown in Fig. 9 (b).

Life-cycle simulation analysis of management risk
In the management risk subsystem, the b 7 factor appears after the planning stage, so it is constrained by a time function.
In order to further explore the dynamic impact of each factor on the management risk subsystem, the initial value of each factor was reduced by 30% and other factors remained unchanged for sensitivity analysis. The adjustment scheme is shown in Table 5, and the nal analysis result is shown in Fig. 9 (c).
Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js Sensitivity analysis shows that regulatory risk can be effectively reduced by controlling regulatory risk at the initial stage of start-up. In the category of management risk subsystem, management risk is most sensitive to b 1 factor, followed by b 4 and b 2 , indicating that perfecting laws and regulations and performance appraisal mechanism, strengthening government attention and establishing overall management organization have obvious effect on reducing management risk in the start-up stage.

Life-cycle simulation analysis of capital risk
In the subsystem of capital risk, only the c 1 factor exists in all stages of the whole life cycle, while the c 4 factor only affects the planning and design stage, the c 3 factor only affects the construction stage, and the c 4 factor only appears in the operation and maintenance stage. Therefore, the time function is used as the constraint for these three factors. In order to further explore the dynamic impact of each factor on the capital risk system, the initial value of each factor was reduced by 30% and other factors remained unchanged for sensitivity analysis. The adjustment scheme is shown in Table 6, and the nal analysis result is shown in Fig. 9 (d). Through sensitivity analysis, it can be found that the in uence of regulation, management and social risk control is far greater than that of the internal factors of the fund subsystem. Among them, management risk has a particularly signi cant impact on capital risk, indicating that a perfect management mechanism plays an important role in capital investment and allocation. All risk factors under the capital subsystem have little in uence on the system, among which operation and maintenance capital has a relatively high in uence on the capital subsystem.

Full life cycle simulation analysis of technical risk
In the subsystem of technical risk, three sub-factors, d 1 , d 2 and d 3 , respectively exist in the stages of planning, construction and operation and maintenance, all of which are constrained by time function. In order to further explore the in uence of each sub-factor on the technical risk system, the initial value of each factor was reduced by 30% and other factors remained unchanged for sensitivity analysis. The adjustment scheme is shown in Table 7, and the nal analysis result is shown in Fig. 9 (e).

Life-cycle simulation analysis of social risks
In the subsystem of social risk, e 1 , e 2 , e 3 and e 4 factors exist in all stages of the whole life cycle, so there is no need to use time function to constrain them. In order to further explore the impact of each factor on the social risk system, the initial value of each factor was reduced by 30% and other factors remained unchanged for sensitivity analysis. The adjustment scheme is shown in Table 8, and the nal analysis result is shown in Fig. 9 (f). Through sensitivity analysis, it can be found that the control of management risks will have a signi cant impact on social risks, which may be due to the government's emphasis on publicity and education, and the improvement of public participation mechanism, so as to reduce social risks. In the category of social risk subsystem, social risk is sensitive to e 1 and e 3 . With the control of such risks in the early stage of the project, social risks can be effectively reduced.

Life-cycle simulation analysis of sponge city construction risk
In order to analyze the in uence of the sub-systems of legal risk, management risk, capital risk, technical risk and social risk on the total risk of sponge city construction, the initial risk value of each sub-system was reduced by 30%, and other factors remained unchanged, and the dynamic change of the total risk level was observed. The adjustment scheme is shown in Table 9, and the nal analysis result is shown in Fig. 9 (g). As can be seen from Fig. 9 (g), the risk level of management risk in all stages of the whole life cycle is high, and the management system needs to be further improved in the future. The risk level of legal risk in the start-up stage is high, but with the change of time, the proportion of the total risk of sponge city construction is getting lower and lower. In contrast, technological and social risks have become more and more important over time. The risk level of capital risk in all stages of the whole life cycle is obviously lower than that of other systems, which may be because jiangxi province has a high level of economic development, and the support of national and provincial pilot cities makes local governments have low economic worries in sponge city construction. Therefore, management risk and technical risk should be focused on in the whole life cycle stage, regulatory risk control should be strengthened in the start-up stage, and social risk control should be strengthened in the operation and maintenance stage.

Life-cycle key risk factors radar map
Will be entirely into the startup phase of mentioned above, planning and design stage, construction stage, completion acceptance and operation maintenance phase, each risk subsystem simulation results in 1 Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js month, 3 months, in September and 15 months in the key points of the summary to the radar map, can be more intuitive to see changes in the different stages of the risk, as shown in Fig. 9 (h). The data of system dynamics constructed in this study are mostly subjective and qualitative, and the simulation accuracy is low, so the precise value of risk level cannot be obtained. However, the dynamic change trend of risk level of each subsystem in the whole life cycle of sponge city can be discussed from the general change trend.

Suggestions on risk prevention and control of sponge city construction
The ve dimensions of risk run through each life cycle stage of sponge city construction, but have different development trends. Therefore, risk management should also dynamically consider the impact of various risks on the overall project and put forward suggestions. Combined with simulation and sensitivity analysis, the following suggestions are proposed for the key factors of the ve subsystems. (1) Sponge city pilot cities should actively carry out the compilation of operation and maintenance speci cations in the future, and further revise them according to the actual project operation situation. (2) in the construction of the sponge, etc in the process of the organization as a whole, we should widely transferring backbone, each related department in order to establish a direct and strong, solid work, coordinate clear in different stages of the urban construction, the main responsibility of the sponge and setting up a long-term mechanism to sponge a gradual shift in urban construction are normalized to the construction of content. (3) Use technical means to build an intelligent scheduling platform for sponge facilities, assist scheduling management with comprehensive information, and realize the whole process comprehensive management of sponge city planning, construction, operation management and environmental performance. (4) Strengthen the publicity of sponge city concept, guide social participation, solicit public opinions, understand public preferences, and incorporate reasonable opinions into sponge city planning. (5) While strengthening sponge city publicity, effectively guide market participation and sustainable investment of capital, fully guide the enthusiasm of social capital to participate in sponge city construction, cultivate and guide the incubation of sponge city-related industries, and form new industrial momentum.

Conclusion
In recent years, China has paid more and more attention to the problems of urban water security, water resources and water ecology.This study nally identi ed 23 risk factors affecting the construction effect of sponge city under the ve sub-systems of regulation, management, capital, technology and society, judged the life cycle stages of the main risks, analyzed the risk causes, and preliminarily concluded the mechanism of action among the ve sub-systems.It is found that management risk and regulatory risk have a signi cant impact on other risk subsystems.Through sensitivity analysis, it is found that a 1 , b 1 , c 4 , d 1 , e 3 and other factors signi cantly affect the effectiveness of sponge city construction.By summarizing and analyzing the risk changes of the whole life cycle through the radar chart, it is found that the Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js management risk has a great impact on each stage of the whole life cycle, the technical and regulatory risks continue to climb along with the life cycle, while the regulations decline, and the level of capital risk is always low.
There are also some shortcomings in this study. In the selection of indicators, the indicators are mostly subjective factors such as "poor technology" and "imperfect system and mechanism", which are directly determined by assigning values through expert interviews, lacking rigorous theoretical veri cation.Although it does not have a signi cant in uence on the running state of the system in the simulation of system dynamics model.However, more scienti c methods should be used to determine the values of constants and phase relation in the model in the future.

Declarations
Ethical Approval There are no ethical issues with this study.  Schematic diagram of sponge city action mechanism (The picture shows the construction composition and water resource utilization principle of sponge city) Figure 4