(1)
Cheng Xiaotao, Qiu Jingwei, Li Na and Liu Shukun
China Institute of Water Resources and Hydropower Research (IWHR)
A-1 Fuxing Road, Beijing, 100038, China
Tel: +86-10-68515511,ext.1595, Facsimile +86-10-68538685, Email: chengxt@iwhr.com
Abstract：China is now on the way of rapid urbanization. Especially, the cities in coastal region and by the major rivers are developing in high speed and facing distinct changes of flood risks. In order to hold variability of flood risk in urbanized areas, urban flood simulation techniques have been developed by IWHR in the last 10 years. In this paper, the progress of urban flood simulation techniques in China is reviewed, and their features are illustrated, taking some case studies of Shenzhen, Guangzhou and Tianjin as examples.
Keywords:
urban flood control, flood risk analysis, urban
flood simulation
China is now on the way of rapid urbanization. There are 95% of the existing 672 cities suffered from flood hazards. Especially, the cities in coastal region or by the major rivers are developing in high speed and facing distinct changes of flood risks. The urbanized areas have been varying obviously in landform and topography, going with increased density of buildings, expanded areas with impervious blanket, developed engineering systems for flood control and storm drainage, and more embankments of railways and highways. Since the underlying conditions for rainfall-runoff and flood routing have been changed, the statistical regular patterns based on the historical flood records are unable to represent the features of urban flood risk distribution at present and in future.
The Urban Flood Simulation Model (UFSM) developed by IWHR is able to simulate the rainfall-runoff and flood routing phenomena numerically, basing on the theory of 1D and 2D unsteady flow, and absorbing the advantages of hydrology. The changes of underlying conditions in urban area can be considered well. In the last 10 years, such techniques developed quickly and utilized in some large cities as Shenyang, Guangzhou, Haikou, Shenzhen, Shanghai, Tianjin and Haerbin.
2D unsteady flow hydraulic numerical models were developed to calculate flow in bay, lake and river course in their early stage. Iwasa, Y. and Inoue, K. (1980) developed a numerical model properly to simulate flooding with shape moving boundary phenomena. In the middle of 1980s, Liu, S. improved the model and applied it in China, especially for the detention areas. In the end of 1980s, the model started to be used in urban areas, the effects of water sluices, pump stations, culverts, bridges, weirs, as well as density of buildings, were considered in the models of Shenyang and Guangzhou cities. For the buildings, it was supposed that water might enter the building at certain level of inundation to take its impacts on the balance of water volumes into account (Liu, S. etc, 1987&1993). In 1993, the regular grid model was improved for flood risk assessment to the flood control planning in Shenzhen City. The hydrological means was fetched into the hydraulic model to simulate the rainfall-runoff well. A set of standard blocks tumbling 1D canal into 2D grid was designed to represent the smaller channel systems. A pre- and post-data treatment system and the visible function were developed for the model (X. Cheng & J. Qiu, 1995). Since 1995, a 2D model with non-structural irregular grid has been utilized into urban areas. New functions to simulate sewer systems and special road-type grids have been developed. From 1998, cooperated with the Tianjin Institute of Meteorology Research, the model has been further improved to become a real-time forecasting model taking the on-line rainstorm measured and predicted information as the input boundary condition. Both the model and its information system reformed to run in the WINDOWS circumstance (J. Qiu, X. Cheng & N. Li, 1999).
Development of urban flood simulation model in China faces many troubles, such as big area, long period of flooding, large scale and insufficient available data. Firstly, we should consider how to take the essential factors and simplify the object in large scale reasonably under restrictive conditions. The second is how to advance efficiency in making, adjusting and running the model and in outcome analysis for such large and complicated model. In this paper, Shenzhen, Guangzhou and Tianjin models will be illustrated as examples to show how to solve the problems.
Shenzhen is a developing city, located in the South China just near Hong Kong. Shenzhen River basin is a small area of 321.25km2, in which Shenzhen Reservoir with controlled area of 60.5km2 and Honghu Detention Area with storage of 2,500,000 m3 were constructed, respectively. In 1993 and 1994, the city was struck by excessive rainfall and flooded three times. Then, the planning of urban flood control engineering system was intensified. However, due to the rapid urbanization, the inherent rainfall-runoff features have been changing obviously. Therefore, a numerical model was required to offer the basis for reasonable assessment of the planning. A 2D model was developed to cover the whole Shenzhen River basin including some northern part of Hong Kong. The governing equations of 2D unsteady flow are adapted as follows:
Continuous equation:
(1)
Momentum equation:
(2)
(3)
Where, H is water depth, Z is water level, q is source item, which consists q0 , the effective rainfall strength and q1, drainage strength of underground drainage network. M, N are the unit width discharges and u, v are the components of flow speed in the x and y directions. n is the coefficient of roughness. g is the acceleration due to gravity.
In the model, steps of grid are taken as 500m5500m. In order to take the effects of narrow stream or drainage channels, the method of ditching in regular 2D grids was developed, and flow in trenches were calculated by 1D unsteady flow method:
(4)
Where, Q is the discharge, A is wetted cross-sectional area, and Sl is friction slope.
A one/two-dimensional mixed method is developed, i.e., for each trenched grid, two elevations and two water depths are admitted, keeping only one water level which may be lower than ground level. The grids with trenches can be simplified to the standard patterns (Fig.1), by which the drainage channel system in the calculation region can be composed just like playing toy bricks (Fig. 2).
Fig.1 Classification of special meshes in 1D and 2D composite model
Fig. 2 Layout of meshes in 1-D & 2-D composite model
The model has been verified by the measured flood data of September 26, 1993. The calculated submerging scope (for those grids with water depth higher than 0.05m) coincides with the observed scope well. The water stage hydrograph measured at two points during the flood were used to compare with the calculated results (Fig.3 & Fig.4).
Fig.3 Verification of water level in front of the Honghu Sluice for the flood of Sept.26, 1993
Fig.4 Verification of water level in front of
the water gate of
Shenzhen Reserves for the flood of Sept.26, 1993
Passing through the verification and adjustment, the model was applied to: ¬understand the features of flood risk distribution in Shenzhen City; evaluate the effects of structural measures in the urban flood control planning. Calculation results show that for the flood caused by the 50-year and the 100-year rainstorm condition, the inundated area will be decreased obviously. However, for the Luohu District, the situation would get even worse because the new dike may hamper storm water into the channel. Some new pump stations should be added over there; ®analyze the effects of further urbanization on flood hazards in Shenzhen City by comparing with six different schemes. For example, it was supposed in Scheme I that the rates of impermeable area in existing urbanized region were increased from the average of 45% to 90%. In this case, the inundated water depth might be risen with increments of 0.1~0.5m., see Fig. 5 and Fig.6.
Fig.5 Comparison of the inundated water depth process at Worker’s Cultural Center in Scheme I
Fig.6 Comparison of the inundated water depth process at Postal Service Building in Scheme I
Since 1995, the original model has been transformed significantly by using of nonstructural and irregular grid techniques to reflect the impacts of complex topography and engineering measures on flooding process in Urban Flood Simulation Model. The new model, taking Guangzhou City as its object, was developed in the Key Program of Urban Flood Strategy Research supported by the National Fund of Natural Sciences. The nonstructural and irregular grid model was originally developed during the 8th 5-year National Key Scientific Research Program (1993-1995), in light of the characteristics of the floodway and detention areas on the Lower Yellow River. (1) The new model keeps the major merits of regular grid model. The water elevation is calculated on mesh center, and discharges are calculated through each cell interface, while the meshes can be triangle, 4-sides or 5-sides; the direction of normal line of links can be free, see Fig. 7. (2) The blocking structures such as dikes or levees can be designed on the links with their actual direction; wide rivers or rivers with dike can be simplified into the river-type mesh; narrow rivers or channel can still be set inside of meshes as shown in Fig. 8. All of these make simplified topography approach to the real situation. (3) Because of the nonstructural feature of the model, it is possible to adjust it by moving nodes and adding or deleting meshes to adopt topography changes better.
Fig.7 Discrete layout of the basic state variables
Fig.8 Meshes layout in 1D & 2D composite model with irregular grids
The calculation area of Guangzhou City is 468m2 (Fig.9). It is clear that the feature of rivers and topography are able to reflect better with irregular grid. Because calcula-tion area is large and simulated objects are complicated, the key of modeling lies in how to rationally simplify the simulated objects at most to meet the demands of calculation precision. Thus, the major factors should be grasped to make the model as simple as possible.
Fig.9 Flood simulation model with irregular grids fro Guangzhou
In the simplified model, the water volume balance is fully satisfied in solving Continuity Equation (1) based on the Finite Volume approach:
(5)
where:
is the unit discharge through the kth interface of the ith mesh;
is the length of the kth interface
of the ith mesh.
For irregular grid, if a little change of water level is assumed on each mash in the period of DT, the unit discharge through each interface of any mash can be obtained by using simplified classification-processing methods
(1) River type: they are designed between river type meshes. The local acceleration term, gravitation term and resistance term are retained in Momentum Equation with a discrete form:
(6)
Where: hj is the mean water depth at jth passage. If the slope is less than 0.0006, hj is taken as the average of water depths in adjacent meshes; otherwise, it is treated as a stepped form. DLj is the sum of distances from the center point of a interface to the centroids of two adjacent meshes.
(2) Road type: Roads in urban area are usually flat, and flood routing is mainly influenced by gravity and resistance. After neglecting the item of acceleration, the following equation can be obtained.
(7)
Where: DL is the centroidal distances between two adjacent meshes; Sign is the judgment function of (+/-)
(3) Others: a) Water retarding structure’s type: such as dikes, embankments of highway and railways, etc, are summarized to continuous levees or levees with breach, the Q is calculated by weir-flow formulae. b) Bridge, culvert and sluice type: such as the gated outlets of lakes, the Q is calculated by weir-flow formulae in free flow conditions, or by orifice-flow formula in pressure flow conditions. c) Pumping type: the water depth in mesh centroid is used as the judging criteria. When water depth exceeds the specified value, the Q is determined by the drainage-capacity of the pump station. d) Open drainage channel type, containing two cases: open channel at both sides of the interface or at just one side. Flows in the channel and over the ground are calculated separately.
Application of the UFSM involves in large number of pre-process for basic information and post-process and analysis of calculation results. A completed pre-process and post-process information system is developed to enhance the efficiency. The structure of UFSM is showed in Fig. 10.
In 1998, in order to propel the National Flood Risk Map Spreading Program, RCDE took a pilot study projects of flood risk mapping. One of its contents is dike-break risk research for Beijiang Flood Protection Area in Guangdong Province. Beijiang Dike is an important flood protective screen for Guangzhou City. It was broken at several places during the great flooding of 1915, which made Guangzhou City inundated for 7 days. The technology scheme of this project is to combine the flood simulation model with GIS. The model is used to calculate the flooding overflow process and reflect the temporal-spatial features of flood risk in current situation; GIS is used to describe various risk information by a set of digital maps. It is to probe the suitable displaying form of flood risk maps, which may provide strong technical supports for flood risk management by flood control agencies in China.
Fig.10 Structural diagram of the urban flood simulation model
Since the nonstructural irregular gird is expandable, it is possible to adjust model calculation area based on the urban expanding or the demanding of other projects. Therefore, the area of existed model for Guangzhou City has been enlarged to the total area of Beijiang Flood Protection Area, the calculation area is up to 3000 km2 (Fig. 11). The New model has also been transplanted from DOS system to Windows system, and combined with MapInfo that make the calculation results analysis and displaying more vivid (see Fig. 12).
Fig.11 Beijiang Dike-break risk assessment model
Fig.
12 Beijiang dike-break flood risk zoning map
In the process of urbanization, owing to the effect of urban hot-island, increase of impermeable area, subsidence of ground level, and so on, the capability of urban drainage system is relative insufficient and the flooding in urban area is tending to deteriorate. Tianjin City, which lies in the lower reach of the Haihe River basin with lower terrain, is such a typical example.
Since 1998, RCDE has cooperated with Tianjin Meteorological Research Department (TMRD) of Tianjin Meteorological Bureau to develop a Rainstorm and Flood Simulation System for Tianjin City. Based on the previous UFSM, some new functions were added to the Tianjin model, such as rapidly processing real-time rainfall data and numerical weather forecast data. These functions provide real-time and uneven distribution of rainfall as boundary conditions of UFSM. Thus, the flood simulation model is combined with a Heavy Rainfall Prediction System, creating a Rainstorm and Flood Forecasting and Early Warning System for Tianjin City (Fig. 13).
Fig.13 Grids map of flood simulation model for Tianjin
The simulation method of urban drainage system is improved in the Tianjin model. Since only the major pipe distribution and pipe scales are known, the pipe system have to be simplified in each mesh according to the mesh area, elevation, density of buildings, the trend of the pipes, pump capacity in different divisions, etc. The rationality of the simplified drainage system is ensured by contrasting the calculation results with observed data of rainfall and inundation on Aug. 16, 1995 and Apr. 23, 1998 in Tianjin City. For narrow open channels with dike on both sides of them, a special node-type grid has been developed, see Fig.14 and Fig. 15.
Fig.14 Diagram of special passage calculation
Fig.15 Layout of drainage
system
The Urban Flood Simulation System (UFSS) for Tianjin City is composed of 4 components: U/C interface, data processing, and model operating and file management. The information pre-processing functions include meshes dividing, graph displaying, and information picking-up, edit, revision, saving, query and error correcting. It can also process rainfall monitor data and real-time prediction data, inserting them into mesh center. The information post-processing function includes the displaying of floods process, statistical analysis and displaying the maximum inundation area, maximum water depth, the query of maximum water depth or water depth process at any mesh and important area, etc.
The system is developed upon Chinese Windows 98, using Visual Basic 5.0 as the main language for system development, and Power Station 4.0 for simulation model development. Since all the data processing and graph displaying functions are developed through self-reliance, UFSS is powerful in its specialties, convenient in utilization, and the cost of modeling is much lower than choosing commercial software.
The Urban flood numerical simulation technology has developed rapidly in China in recent 10 years, which has been applied in the cities of Shenyang, Haikou, Guangzhou, Shenzhen, Harbin, Tianjin and Shanghai. It has been improved step by step into the stage of its practicability. In future, its simulation functions should be improved in operating the flood control works in UFSS, using real-time observed data to correct the calculation results, strengthening database management, and developing information update technology from multi-information-source, etc.
The development of UFSS was greatly supported by the Water Bureau of Shenzhen City, Guangzhou Water Conservancy & Electric Power Bureau and Tianjin Meteorological Research Department. The authors are grateful to Prof. Lu Jikang, Chen Hao, Chen Xijun, Lan Hong, Yang Lei, Guan Jianyong at RCDE for their contributions to UFSS.
References
Iwasa Y. and Inoue K., Numerical Method for Inundating Flood, Annals of Disaster Prevention Research Institute of the Kyoto University, No. 23, B-2, 1980 (in Japanese).
Liu Shukun and Yu Tianyi, et al: Numerical Simulation of Flood Routing over Floodplain and Flood Risk Map, Research Report of IWHR, 1987 (in Chinese).
Liu Shukun, Du Yi, Fu Zengchi and Zhou Kuiyi: Guidebook for Flood Control and Disaster Mitigation, Liaoning People’s Press, 1993 (in Chinese).
Cheng Xiaotao & Qiu Jingwei, et al : Numerical Simulation and Analysis of Flood Hazard and Disaster Mitigation for Shenzhen City, Journal of Natural Disasters, Vol 4, Supplementary issue, 1995 (in Chinese).
Cheng Xiaotao & Qiu Jingwei. et al: Research on the Urban Flood Simulation System, Journal of China Institute of Water Resources and Hydropower Research, Vol.1, No.2, 1997 (in Chinese).
Qiu Jingwei, Cheng Xiaotao & Li Na et al: Development of the Rainstorm and Flood Simulation System for Tianjin City, Proceeding of the 2nd National Urban Water Resources Symposium, 1999 (in Chinese).
Chen Hao, Qiu Jingwei et al: Pilot Study of Flood Risk Map-making: Flood Protected Region of Beijiang Dike, Research Report of China Institute of Water Resources and Hydropower Research, 1999 (in Chinese).
Urban River Division: Handbook of Flood Simulation and Verification of New Model, Report of the Public Works Research Institute No. 3400, Feb. 1996.