Dam-Break Flood Routing Simulation and Scale Effect Analysis Based on Virtual Geographic Environment

Dam-Break Flood Routing Simulation and Scale Effect Analysis Based on Virtual

Geographic Environment

Jun Zhu,Lingzhi Yin,Jinhong Wang,Heng Zhang,Ya Hu,and Zhujun Liu

Abstract—Because of the barrier lake affected by highly changing and complex environment,it is always a great challenge that proper actions must be performed within a limited amount of time.Thus,how to scienti?cally and ef?ciently analyze the dam-break risk of the barrier lake and its impact area is very important for barrier lake disposal and downstream communities transfer.To improve the ef?ciency and the accuracy of the dam-break?ood routing simulation,this paper mainly focuses on the construction of the virtual geographic environment(VGE)system and the scale effect analysis.Unlike most of the current cellular automata(CA)-based?ood simulation systems,the proposed VGE system can offer an intuitive,ef?cient,and interactive visualization environment through which users can explore com-plicated spatial information and conduct risk assessment work. Some key technologies including the CA model of dam-break ?ood,the VGE system framework,the impact analysis method, and the scale effect analysis method were discussed in detail.A prototype system was developed to support dam-break simulation and risk analysis of the Xiaojiaoqiao barrier lake in Anxian County,Sichuan Province of China.By means of a variety of cell scales effect analysis experiments,the adaptation scope and characteristics of multiscale cells were obtained to implement the simulation analysis of the dam-break?ood routing better. The proposed VGE system can improve the ef?ciency of risk assessment and decision-making.

Index Terms—Cellular automata(CA),dam-break?ood,mul-tiscale,risk assessment,simulation analysis,virtual geographic environment(VGE).

I.I NTRODUCTION

F LOOD AND landslides are among the most widespread

natural hazards on the earth[1].Dam-break?ood is one of them.The dam-break disaster brings hazard risk,including human injury and death,properties loss,and infrastructure destruction[2].It presents serious threats to people and properties due to possible upstream?ooding with the dammed Manuscript received November15,2013;revised June05,2014; accepted July13,2014.Date of publication August10,2014;date of current version February04,2015.This work was supported in part by the National Natural Science Foundation of China under Grant41001252and Grant41271389,in part by the Open Research Fund by Sichuan Emergency Mapping Support and Geological Disaster Monitoring Engineering Research Center under Grant K2014B007,and in part by the Graduate Innovation Fund of Southwest Jiaotong University under Grant YC201414233.(Corresponding author:Lingzhi Yin.)

The authors are with the Department of Remote Sensing and GeoSpatial Information Engineering,Faculty of Geosciences and Environmental Engi-neering,Southwest Jiaotong University,Chengdu611756,China(e-mail: ylz9001@https://www.360docs.net/doc/319004176.html,).

Color versions of one or more of the?gures in this paper are available online at https://www.360docs.net/doc/319004176.html,.

Digital Object Identi?er10.1109/JSTARS.2014.2340399lake water level raise and possible downstream?ooding due to dam breach and rapid release of the dammed water[3].The dam-break?ood simulation can predict the spread of the dam-break?ood process,provide information such as water depth,?ow velocity,etc.,and determine the inundation areas and disaster degree.It achieves rapid estimation of dam damage simultaneously,the loss due to the?ood disaster,and provides necessary information for emergency rescue plan.It can also provide necessary information,such as early warning of the disaster.So,it is very helpful for preventing disaster,protecting people’s life,and also useful for designing the height of dam (or cofferdam)and drainage of standard building. However,dam-break?ood is very sensitive to highly chang-ing and complex environment,which has some characteristics such as sudden occurrence,rapid expansion,and urgency respond[4].Thus,it is quite dif?cult to simulate dam-break ?ood.The cellular automata(CA)is a dynamic discrete space and time system,which offers a framework for the exploration of complex adaptive systems[5].Li et al.[6],[7],Dottori and Todini[8],and Gallegos et al.[9]have used the CA model to implement the dam-break?ood routing.However,there are some shortcomings of these studies:

First,the CA model is sensitive to different scales.However, the existing CA-based hydrological and hydraulic models only used a single regular grid to divide the cellular https://www.360docs.net/doc/319004176.html,cking of cell-scale sensitive analysis may result in inaccurate results. Second,because the calculation of dam-break?ood based on high-resolution DEM is time-consuming,it is dif?cult to meet the requirement of“real-time”simulation analysis of the dam-break?ood routing.Hence,how to choose a balance between the accuracy and the ef?ciency of the dam-break?ood routing is particularly important.

Finally,multisource heterogeneous data are dif?cult to tightly couple.Furthermore,the model calculation,the risk assessment,and the decision-making are also separated.It is urgent that a highly ef?cient geographic simulation and analysis tool can be offered to support the rapid dam-break risk prediction and decision-making.

The virtual geographic environment(VGE)was proposed to offer an intuitive,ef?cient,and interactive visualization environment through which geographically separated users can explore complicated spatial information and conduct collab-orative work[10].Compared to traditional GIS,the VGE pays more attention to multisource data integration,sharing, and information mining[10].Using the geographic analysis

1939-1404?2014IEEE.Personal use is permitted,but republication/redistribution requires IEEE permission.

See https://www.360docs.net/doc/319004176.html,/publications standards/publications/rights/index.html for more information.

model and various expression technologies,the VGE is used to achieve geographic problem analysis,geographic rule ex-traction,geographic phenomena simulation,geographical rep-resentation and predict environmental changes,and human activities impact assessment[10],[11].Thus,the VGE cannot only be used as a tool for geo-object-based multidimensional spatial analysis and multichannel geo-interaction,but can also be used as a platform for geo-process-based simulation of dynamic geographic phenomena.

This paper focuses on the construction of the VGE and the effect analysis of the multiscale cell,which aims to improve the simulation accuracy and the analysis ef?ciency of the barrier lake risk assessment.This paper is organized as follows.In Section II,the features of the CA model and the VGE technologies are introduced and their advantages are discussed.In Section III,several key techniques,such as the CA model construction,impact analysis method of the dam-break?ood,and the scale effect analysis,are discussed in detail.In Section IV,a prototype system is developed and rudimentary experimentation is implemented.Finally,conclu-sion and discussion are addressed in Section V.

II.B ACKGROUND

The CA model is a dynamic discrete space and time system [5].It has the ability to simulate complicated geographical phenomenon from the bottom up.It also provides a new perspective to explore the pattern,process,and evolution of geographical phenomenon.The CA model is widely used in urban evolution,land use,traf?c simulation,and other ?elds[12]–[14].In?ood-related simulations,Li et al.[6], [7]proposed a CA model to estimate?ood loss from the barrier lake by incorporating natural and social factor and did the sensitivity analysis concerning several physical pa-rameters,such as manning’s coef?cient and in?ltration values and population density[5],[6].Dottori and Todini proposed a two-dimensional(2-D)hydraulic model based on the CA approach,which has less runtime and data requirements,and maintains the accuracy and reliability of the simulation process [8].Gallegos et al.put forward an unstructured grid,Godunov-type,?nite volume model that solves the shallow-water equa-tions and high-precision DTM’s outburst?ood model based on BreZo,which was applied to southern California City of residence[9].

However,the spatial scale decisions in the above CA-based hydrological and hydraulic model are often made arbitrarily or in relation to data availability.For the complex system modeling using a CA approach,scale is an important concept. The CA model is sensitive to the cell size.M′e nard and Marceau evaluated the sensitivity of the land-cover change model to spatial scale and found that the spatial scale has a considerable impact on simulation dynamics in terms of both land-cover area and spatial structure[16].Samat investigated the effect of changing scale on the performance of a GIS-based CA model developed to simulate the spatial pattern of urban growth,found that the model performed well and produced realistic urban form only in a speci?c range of scale[15]. Kocabas and Dragicevic explored the impact of neighbourhood size and type on the outcomes of a GIS-based CA urban growth model and found that the CA neighbourhood size and type con?gurations had a signi?cant in?uence on the CA model output[18].Pan et al.explored the impact of variation in scale on the behavior of a CA employed for land use change modeling and found that the variation of the spatial extent,cell size,and neighborhood size and shape in?uenced the model’s behavior in different ways[17].

Moreover,the above researches usually focused on how to construct a CA model,and pay little attention to the integrated study of model calculation,risk assessment,and decision-making,which may result in lacking of the practical appli-cation.VGE is proposed as a new generation of geographic analysis tool to contribute to human understanding of the geographic world and assist in solving geographic problems at a deeper level[10].Three use cases are identi?ed that together encompass the current state of the VGE at different application levels:1)a tool for geo-object-based multidimensional spatial analysis and multichannel interaction;2)a platform for geo-process-based simulation of dynamic geographic phenomena; and3)a workspace for multiparticipant-based collaborative geographic experiments[10].Creating virtual environments have great potential for research in simulation and analysis of complex geographic data and phenomena[19],[20].There are a lot of the VGE application paradigms.Zhu et https://www.360docs.net/doc/319004176.html,ed the VGE platform to implement?ood peak forecasting[21]. Xu et https://www.360docs.net/doc/319004176.html,ed the VGE platform to evaluate the air pollution and improved the ef?ciency of computation and evaluation [22].Lin and Zhu used the CVGE platform to plan and design silt dam system and improve the ef?ciency[23],[24]. Gong et https://www.360docs.net/doc/319004176.html,ed the VGE platform to simulate the SARS transmission[25].

Therefore,this paper will build a VGE platform for dam-break?ood simulation and disaster assessment.It can integrate multisource heterogeneous data and spatial analysis model to provide an intuitive,ef?cient,and interactive visualization environment for users.On the basis of VGE platform,we will generate multiscale cell to carry out dam-break?ood simulation experiments,and simulation results in different cell sizes will be discussed in order to?nd out the proper effect in different applications.This paper mainly focuses on how to solve the following four problems:1)the CA model construction;2)the VGE platform construction;3)the impact analysis of the dam-break?ood;and4)the scale effect analysis method.

III.M ETHODOLOGY

The dam-break?ood simulation can predict the spread pro-cess of the dam-break?ood,provide information such as water depth and?ow velocity,and determine the inundation areas and disaster degree.This section will introduce the method of dam-break simulation in the VGE platform,including the CA model of dam-break?ood,the framework of the VGE system, the impact analysis method,and the scale effect analysis method of the dam-break?ood.

A.CA Model Construction

A classic CA system consists of a regular grid of cells, each of which can be in one of a?nite number of k possible

ZHU et al.:DAM-BREAK FLOOD ROUTING SIMULATION AND SCALE EFFECT ANALYSIS107 states,updated synchronously in discrete time steps according

to a local identical interaction rule[5].A CA model is thus

composed of four principle elements:1)a lattice;2)a set of

allowed states;3)neighborhoods de?ned by the lattice;and4)

transition rules[26].

1)CA State De?nition:According to the calculation

method of dam-break?ood routing and the existing observa-

tion data,the evolution of each cell is determined by the?ow

velocity and water level of the river source.Hence,each cell

should contain?ve properties:1)the river bed height(DEM);

2)the roughness;3)the water depth,4)the single-width?ux

in x and5)the single-width?ux in y direction.

2)Cellular Space and Neighborhood:This paper uses a

2-D spatial distribution CA model and the study area is divided

into discrete grid according to a certain scale.Grid cell is a

square unit cell with a?xed evolving time intervalΔt.The

model uses the V on Nuemann neighborhood type,uses the

four neighboring grid as a cellular space simultaneously.

3)Transition Rules:Dam-break?ood routing process is

calculated in two steps.The?rst step calculates the single-

width?uxes at time t through the water depth at time t.

The equations come from Saint–Venant Equations,which are

discretized by Li et al.[7]and?t to the CA framework.It is

shown as

????????????

???????????M t+1

i,j

=M t i,j?gΔt(h t i+1,j+h t i,j)(z t i+1,j?z t i,j)

Δx

?gn2i,jˉu i,jΔt

(u t i,j)2+(v t i,j)2

[(h t i+1,j+h t i,j)/2]1/3

N t+1

i,j

=N t i,j?gΔt(h t i+1,j+h t i,j)(z t i+1,j?z t i,j)

Δy

?gn2i,jˉv i,jΔt

(u t i,j)2+(v t i,j)2

[(h t i+1,j+h t i,j)/2]1/3

.(1)

The second step calculates the water depth at time t+1 through the single-width?uxes at time t.The equation is shown as

h t+1 i,j =h t i,j?

Δt(M t+1

i+1,j

?M t+1

i,j

)

Δx

?

Δt(N t+1

i,j+1

?N t+1

i,j

)

Δy

.

(2)

In the above equations,M t i,j and N t i,j are the single-width?uxes(m2/s)of CA cell(i,j)at time t in x and y directions;h t i,j represents the average water depth(m)of CA cell(i,j)at time t;u t i,j and v t i,j represent the average horizontal water speed(m/s)of CA cell(i,j)at time t in the x and y directions,respectively;n2i,j is the hydraulic roughness coef?cient(m?1/3s)of CA cell(i,j);andΔx,Δy, andΔt represent the temporal unit(s),the spatial unit in the x direction(m),and the spatial unit(m)in the y direction, respectively.So,according to the movement rule of?ood,if there is water?owing from the upstream,the water depth and velocity can be computed through the transformation rule.Or otherwise,the current cell will become dry.

B.VGE System Framework

The VGE system is designed for dam-break?ood simulation and assessment.Its framework is divided into?ve layers:1) hardware layer;2)the data layer;3)the core service

layer;Fig.1.Framework of the VGE system.

4)the multidimensional presentation layer;and5)the user working layer(see Fig.1).

The hardware support layer is the software and hardware technical support environment of the VGE system.It is the physical carrier and premise of system implementation.

The data layer includes a variety of high-resolution digital elevation model and remote sensing image data,thematic map data such as population and traf?c information,hydrological rainfall and other historical information.It is used to realize the VGE service system data integration and sharing,which is the foundation of the system.

The core service layer is the core of the whole system, and provides the basic professional services.It provides some models such as the spatial analysis function,the?ood inun-dation analysis model,the?ood routing model,and database interface for supporting carrying out risk assessment services. The multidimensional presentation layer provides multi-dimensional perception of virtual geographic scene,shows a variety of situations of dam-break?ood routing simulation,in order to improve system’s ef?ciency of spatial cognition. The user working layer directly faces to users.The user can query,display,analyze,and do other operations in a3-D visualization way through the interface windows to obtain the dam-break thematic maps and tables,as well as other risk assessment results.

C.Impact Analysis of the Dam-Break Flood

For a risk analysis of dam-break?ood,the?ood arrival,the inundation area,the?ood velocity,etc.,are the most important parameters.The disaster then can be estimated based on these parameters.This section will introduce the compute methods of?ood arrival time,the inundation area,and the disaster assessment model.

1)Flood Arrival Time:The time scale of the geographical CA model is the correspondence between CA iteration step length and the time of the real world.Hence,the?ood routing time is determined by the iteration step length of the dam-break?ood model and sensitive to the spatial scale. In this situation,different cell sizes are used to simulate the

108IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERV ATIONS AND REMOTE SENSING,VOL.8,NO.1,JANUARY2015

dam-break?ood in the VGE platform and record the arrival time and analyze it.

2)Inundation Area:The VGE platform can provide the inundation scope and area in real time.Hence,different scales are used to simulate the dam-break?ood in the VGE platform and acquire the inundation scope,in order to analyze it through set operation.The equation is shown as

Difference=(A∪B)?(A∩B)(3) where A is the inundation scope generated by a?ne scale and B is the inundation scope generated by a coarse scale.

The above equation allows to do an evaluation for the increases or decreases of the inundation scope,which is caused by the change in scale.

3)Disaster Assessment:The purpose of dam-break?ood simulation is to make a risk assessment.The risk assessment model is shown in(4)[27],which is used for evaluating the affected buildings.Li et al.[7]changed these model’s parameters for dam-break disaster assessment

FI=0.45WD+0.22WR+0.22V+0.11D(4) where WD is water depth value,WR is water rising velocity, V is the water velocity,and D represents the duration time of dam-break?ood.All values are set to a binary value(0,1) according to different threshold values.For the water velocity, the threshold value is2m/s.The threshold of the duration time is12h[7],[28],[29].The water rising velocity is de?ned as a slow or rapid,its threshold of2m/h.The calculation equation of water depth is shown below,which is proposed by Waarts [30]and Jonkman et al.[31]

WD=0.665×10?3e1.16h max,W D≤1(5) where W D represents the water depth value and h max rep-resents the evolution of the?ood cell corresponding to a maximum depth of each cell during the?ood routing progress.

D.Method of Scale Effect Analysis

Because the CA model is scale-sensitive,different cell sizes and neighborhood settings will obviously affect the simulation results.Moreover,the CA space in different cell sizes contains different cell number,which may cause a big impact on the computational ef?ciency of the dam-break model.Hence, different cell sizes should be used to simulate the dam-break ?ood routing,and choose a proper cell size according to the emergency response needs.

1)Process of Scale Effect Analysis:Fig.2is the?owchart of dam-break?ood simulation progress,including three parts: 1)the data preparation;2)the routing experiment;and3)the simulation analysis.The data preparation part would collect high-resolution DEM,remote sensing image,and population data.Then some processing would be done on these data such as resampling and spatial correction.Based on the processed data,the data of different spatial scales can be obtained,which is used for subsequent dam-break?ood simulation.The?ood routing experiment will carry out dam-break?ood experiment based on the VGE platform which integrated data from different sources,the CA?ood model,the disaster estima-tion model,and the spatial analysis function.The

simulation Fig.2.Flowchart of scale effect analysis process.

results would be collected for the subsequent analysis.The simulation analysis part will compare the simulation results including the?ood arrival time,the inundation area,and the disaster degree and model ef?ciency,in order to make a scale effect analysis and choose a proper cell size in a certain application.

2)Cell Scale Division and Dam Cell Computation:The neighborhood type of the dam-break model proposed in this paper is V on Nuemann type.Hence,the sizes of cell scales are two-time relationship which can be achieved through the resembling of DEM.The dam-break cell in different scales should process high accuracy,because the similarity should be preserved between the real world and the virtual geographic events.The cell at dam-break can be computed through the linear-calculation of dam-break coordinate and origin coordinate of DEM

col=(x?x )/cellsize

row=T otalRows?(y?y )/cellsize.(6) In the above equation,x is the x-coordinate of dam-break. x is the x-coordinate of the upper left corner of DEM.y is the y-coordinate of the breach.y is the x-coordinate of the upper left corner of DEM.TotalRows is the total number of rows of the DEM.

IV.A PPLICATION E XPERIMENT

A.Case Region and System Development

The Xiaojiaqiao barrier lake(104.26??104.294?, 31.635??31.667?)was chosen as the experiment area, which is7km from Xiaoba town,Anxian County,Sichuan Province of China.The dam’s catchment area is154.81km2, the height of the dam is about56–67m,the length is 260m,and the width is390m.The volume of landslide is approximately2.42×106m3and the maximum of the storage capacity is3.0×107m3.According to the experts, the Xiaojiaoqiao barrier lake is at high risk which is second to Tangjiashan Lake.If the dam breaks down,the?ood will bring different impacts to the areas along the Chaping River, nearly ten towns with over200villages and communities, such as the Xiaoba town,the Sangzao town,and the Anchang town.There are a number of town centers,a river and several roads in?ood-affected areas,which is shown in Fig.3.

ZHU et al.:DAM-BREAK FLOOD ROUTING SIMULATION AND SCALE EFFECT ANALYSIS

109

Fig.3.Case

region.

Fig.4.Snapshot of the interface of the VGE system.

To reduce costs and save time,a 3-D visualization prototype VGE system was developed by Visual Studio https://www.360docs.net/doc/319004176.html,,osgEarth,OpenGL,and OpenSceneGraph.osgEarth and OpenSceneGraph were used to develop the virtual geographic scene,and some online map data such as Google Earth,Google Map,and ArcGIS online were also integrated into the VGE system.Fig.4shows the interface of the prototype system.The system provides a real virtual terrain environment based on high spatial-resolution imagery,as is shown in Fig. 5.It can also simulate the dam-break ?ood on the basis of 3-D https://www.360docs.net/doc/319004176.html,ers can directly and naturally obtain the inundated area and disaster degree,and query the data of water depth and water velocity,etc.,as is shown in Fig.6.

B.Dam-Break Flood Routing Simulation

The DEM data used in this paper come from a 1:10000-scale topographic map,which its spatial resolution is 10m.Its vertical error is 0.5m in ?at ground,1.2m in hilly area,2.5m in mountainous area,and 5m in high mountain area.The length of the dam body is 93m,through which the dam-break ?ood ?ows into the downstream channel.Besides,the river channel is narrow.When the cell size is 80m,the model cannot run normally.Hence,a series of cell sizes were chosen to carry out the ?ood routing experiment,which are 10,20,40,and 60m.The space/time ratio was set to 100whereas the break shape is rectangular and the break

types

Fig.5.Multiscale representations of virtual terrain

scene.

Fig.6.Dam-break ?ood simulation and interaction in the VGE system.

are full and half break.Then the dam-break food routing experiment was carried out in the VGE platform.Fig.7shows the simulation of full break for the Xiaojiaqiao barrier lake in May 31,2008.In the full break situation,the dam-break ?ood reaches XiaoBa town at 33min,while reaches SangZao town at 59min.When dam-break ?ood stabilizes it is 2h and 46min.With GIS-assisted inundation potentials analysis,the simulation results can offer us a lot of risk information.For example,the amount of risk population is 13221,and the risk cities include GuanXin village,ZhongXin village,YunFeng village,XiangXi village,ShangQing village,XiaoBa town,and SanZao town.These impact analysis results can effectively support the risk assessment of the dam-break in the Xiaojiaoqiao barrier lake.

The routing results include the ?ood arrival time,the inundation area and disaster population were collected for subsequence scale effect analysis.C.Model Effect Analysis

1)Arrival Time:Fig.8shows the arrival time of ?ood evolution in different spatial cells sizes to different places.When the cell size is 10–40m in full break case,nothing much has changed for routing time;routing time is longer

110IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERV ATIONS AND REMOTE SENSING,VOL.8,NO.1,JANUARY

2015

Fig.7.Flood routing simulation (5.31full

break).

Fig.8.Flood arrival time of ?ood evolution in different spatial scales.

when the cell size is 60m than that one when cell size is 10–40m.In the case of half break,the routing time is not much different from the cell size which is 10–40m,expect when the cell size is 40m the routing time has jumped;but for the cell size which is 60m,the model based on CA cannot run out correctly.

2)Inundation Area:Fig.9shows comparisons of the inundation area by dam-break ?ood in different spatial scales,and the increasing cell size may result in the inundation area smaller.Fig.10shows comparison of the inundation region affected by the dam ?ood in different cell sizes.The change of the cell size results in the decrease of the inundation region in the mountains.Otherwise,it may result in the increase of the inundation region in the plain.

3)Disaster Estimation:This section mainly analyzes the change of ?ood-exposure index (FI)and disaster population in different cell sizes.FI re?ects the probability of the population death in disaster area.Fig.11shows case region’s villages and towns in different cell sizes’FI.In the case of full break,FI have not much difference in different cell sizes for the dam-break ?ood in upstream such as Wuhu village,Huangyang village,and Xaoba town.However,for the downstream cities such as Shangqing village,Sangzao village,and Xiangxi village,the value of FI is maximum in 20and 40m.In addition,in Xiangxi village,the value of FI in 10m is greater than that in 20and 40m while the value of FI in the

cell

Fig.9.Inundation area in different spatial

scales.

Fig.10.Analysis of inundation area in different spatial

scales.

Fig.11.Value of FI in different spatial scales of full break.

size of 60m in Xiaoba village,Shangqing village,Sangzao village,and Xiangxi village are less than that in other cell size.

Fig.12shows the value of FI in different cell sizes of half break.The values of FI are not much different between

ZHU et al.:DAM-BREAK FLOOD ROUTING SIMULATION AND SCALE EFFECT ANALYSIS

111

Fig.12.Value of FI in different spatial scales of half

break.

Fig.13.Numbers of disaster population.

20and 40m,however,in the cell size of 10m,the values of FI in Shangqin village,Sangzao village,and Shangxi village are less than that in 20and 40m,while when the cell size is 60m,they cannot route normally.

4)Disaster Population:According to FI and population density,the numbers of disaster population can be satiated as shown in Fig.13.In the case of full break,the numbers of the disaster population in 10m cell size,20m cell size,and 40m cell size are approximately equal.However,the cell size of 60m has a less number of disaster populations than any others.Under the condition of half break,the risk population in cell size of 20m is largest,and the risk population in cell size of 10m is least.When the cell size is 60m,the dam-break routing model cannot run normally.

In conclusion,the dam-break model based on CA has a certain degree of scale sensitivity.For the case region selected in this paper,in the case of full break,the model will keep stable in the cell sizes of 10–40m;however,in the cell scale of 60m,the result of dam-break routing has some changes.Furthermore,in the size larger than 80m,the dam-break ?ood routing model cannot run.In the case of half break,the model will keep stable in the cell size of 10–40m,and it cannot run in the cell size of 60m,the value of FI and

the

https://www.360docs.net/doc/319004176.html,parison of compute time in different spatial

scales.

https://www.360docs.net/doc/319004176.html,parison of visual ef?ciency in different spatial scales.

numbers of disaster population are maximum in the cell size of 20m.

D.Model Performance Analysis

The dam-break ?ood model has different computational ef?-ciency with different cell sizes in the ?ood routing experiment.Hence,the model computing time and visualization time of different cell sizes will be compared to balance between the accuracy and the ef?ciency of the dam-break ?ood routing.1)Computation of Dam-Break Flood Routing Model:Fig.14shows the compute time of dam-break ?ood routing model in different cell sizes.In the same cell size,the calculation time will has a small increase with the evolution of the dam-break ?ood.Moreover,in the different cell sizes,the calculation time will proportional decrease with the evolution of the dam-break ?ood.

2)Visual Ef?ciency of Dam-Break Flood Routing:Fig.15shows the analysis of visual ef?ciency of the dam-break ?ood routing in different spatial scales.The visualization time increases with the process of the dam-break ?ood routing in the same cell size.Moreover,the visualization time decreases with the increase of different cell sizes.E.Model Calibration and Validation

Dam-break does not really happen in the case region,so the paper makes a comparison with the simulation results in other paper.After a comparison is made between this paper and paper [7],it turns out that the simulation results of this paper and paper [7]are generally consistent with each other in such aspects as ?ood arriving time,inundation area,and disaster people (see Table I).

112IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERV ATIONS AND REMOTE SENSING,VOL.8,NO.1,JANUARY 2015

TABLE I

C OMPARISON B ETWEEN

D IFFERENT R

ESULTS

TABLE II

A DVANTAGES OF THE CA M ODEL IN

VGE

V.C ONCLUSION AND D ISCUSSION

This paper focuses on the construction of the VGE and the effect analysis of the multiscale cell,which aims to improve the simulation accuracy and the analysis ef?ciency of the risk assessment in the barrier lake.The CA model of dam-break ?ood routing,the framework of the VGE system,the impact analysis methods,and the scale effect analysis method were discussed in detail.On the basis of multiscale cell space division,a case region was chosen for carrying out an exper-iment.Experimental results show that the VGE platform can integrate the remote sensing data,DEM,population data and other data,tightly coupled with the dam-break ?ood model,risk assessment models,the spatial analysis function,etc.It provides an intuitive platform used for data management and decision-making of dam-break ?ood.The users can simulate the dam-break ?ood routing progress and implement a risk assessment in a 3-D visualization way which can improve the decision-making ef?ciency.Table II shows the advantages of the CA model in this paper compared to the CA model in paper [8],[32]and ArcGIS software,respectively.

Different cell sizes were used in the simulation experiments of the dam-break ?ood routing.The dam-break ?ood CA model was discovered that it only can work correctly within a certain cell sizes’range.The model ef?ciency will increase with the bigger of the cell size.Hence,for the case region addressed in this paper,the 10-or 20-m cell size can be used to simulate the dam-break ?ood if there is ?ne DEM data and enough time.Cell size of 40m can be used if there is not ?ne DEM data and inadequate time.Hence,according to different features of different cell sizes,a proper cell size should be selected to implement the simulation analysis of the dam-break ?ood routing in one certain urgency response time.Because of the barrier lake affected by highly changing and complex environment,it is always a great challenge that

proper actions must be performed within a limited amount of time.So the dam-break analysis model not only should implement spatiotemporal process simulation,but also con-sider the spatial decision-making behavior.The multi agent system (MAS)technology will be introduced into the system to improve simulation accuracy and analysis ef?ciency in the dynamic and complex environment.Moreover,it is very important to build a highly ef?cient and ?exible collaborative VGE system for intuitive and ef?cient interactive visualization,which allow distributed users to share resources and models,explore complicated geographical phenomenon and implement collaborative risk assessment.

R EFERENCES

[1]Y .Hong,R.Adler,and J.Verdin,“Use of 21st century satellite remote

sensing technology in natural hazard analysis,”Nat.Hazards ,vol.43,no.2,pp.165–166,2007.

[2]S.Yi,Y .Xie,and X.Zuo,“A risk analysis information model with

uncertainty for ?ood hazard of dam-break,”in Proc.Geoinformat.,2010,pp.1–6.

[3]P.Cui,Y .Y .Zhu,Y .S.Han,X.Q.Chen,and J.Q.Zhuang,“The

12May Wenchuan earthquake-induced landslide lakes:Distribution and preliminary risk evaluation,”Landslides ,vol.6,no.3,pp.209–223,2009.

[4]J.Zhu,Y .Li,and Y .Hu,“A virtual geographic environment for

simulation analysis of dam-break ?ood routing,”Adv.Mater.Res.,vol.463,pp.926–931,2012.

[5]J.Cheng,Modeling Spatial and Temporal Urban Growth .Utrecht,The

Netherlands:Utrecht Univ.,2003.

[6]Y .Li et al.,“Ef?cient dam break ?ood simulation methods for develop-ing a preliminary evacuation plan after the Wenchuan earthquake,”Nat.Hazard Earth Syst.,vol.12,no.1,pp.97–106,2012.

[7]Y .Li et al.,“Spatiotemporal simulation and risk analysis of dam-break

?ooding based on cellular automata,”Int.J.Geogr.Inf.Sci.,vol.27,no.10,pp.2043–2059,2013.

[8] F.Dottori and E.Todini,“A 2D ?ood inundation model based on cellular

automata approach,”in https://www.360docs.net/doc/319004176.html,put.Methods Water Resour .(CMWR),2010,pp.602–609.

[9]H.A.Gallegos,J.E.Schubert,and B.F.Sanders,“Two-dimensional,

high-resolution modeling of urban dam-break ?ooding:A case study of Baldwin Hills,California,”Adv.Water Res.,vol.32,no.8,pp.1323–1335,2009.

[10]H.Lin et al.,“Virtual geographic environments (VGEs):A new gener-ation of geographic analysis tool,”Earth Sci.Rev.,vol.126,pp.74–84,2013.

[11]H.Lin,M.Chen,and G.Lu,“Virtual geographic environment:A

workspace for computer-aided geographic experiments,”Ann.Assoc.Amer.Geogr.,vol.103,no.3,pp.465–482,2013.[12]I.Sant′e ,A.M.Garc′?a,D.Miranda,and R.Crecente,“Cellular automata

models for the simulation of real-world urban processes:A review and analysis,”Landscape Urban Plann.,vol.96,no.2,pp.108–122,2010.[13]Q.S.Yang,X.Li,and X.Shi,“Cellular automata for simulating land use

changes based on support vector machines,”Comput.Geosci.,vol.34,no.6,pp.592–602,2008.

[14]W.F.Fang,L.Z.Yang,and W.C.Fan,“Simulation of bi-direction

pedestrian movement using a cellular automata model,”Phys.A ,vol.321,no.3,pp.633–640,2003.

[15]N.Samat,“Characterizing the scale sensitivity of the cellular automata

simulated urban growth:A case study of the Seberang Perai Region,Penang State,Malaysia,”Comput.Environ.Urban Syst.,vol.30,no.6,pp.905–920,2006.[16] A.M′e nard and D.J.Marceau,“Exploration of spatial scale sensitivity

in geographic cellular automata,”Environ.Plann.B ,vol.32,no.5,pp.693–714,2005.

[17]Y .Pan,A.Roth,Z.Yu,and R.Doluschitz,“The impact of variation in

scale on the behavior of a cellular automata used for land use change modeling,”Comput.Environ.Urban Syst.,vol.34,no.5,pp.400–408,2010.

[18]V .Kocabas and S.Dragicevic,“Assessing cellular automata model

behaviour using a sensitivity analysis approach,”Comput.Environ.Urban Syst.,vol.30,no.6,pp.921–953,2006.

[19]W.S.Bainbridge,“The scienti?c research potential of virtual worlds,”

Science ,vol.317,no.5837,pp.472–476,2007.

ZHU et al.:DAM-BREAK FLOOD ROUTING SIMULATION AND SCALE EFFECT ANALYSIS 113

[20]G.Bell,T.Hey,and A.Szalay,“Beyond the data deluge,”Science ,

vol.323,no.5919,pp.1297–1298,2009.

[21]J.Zhu,C.B.Liu,J.H.Gong,D.J.Wang,and T.Song,“A distributed

computing service for neural networks and its application to ?ood peak forecasting,”in Proc.Int.Conf.Neural Inf.Process.(ICONIP),2006,pp.890–896.

[22] B.L.Xu et al.,“Integration of a computational grid and virtual

geographic environment to facilitate air pollution simulation,”Comput.Geosci.,vol.54,pp.184–195,2013.

[23]H.Lin,J.Zhu,J.H.Gong, B.L.Xu,and H.Qi,“A grid-based

collaborative virtual geographic environment for the planning of silt dam systems,”Int.J.Geogr.Inf.Sci.,vol.24,no.4,pp.607–621,2010.[24]J.Zhu,J.H.Gong,W.G.Liu,T.Song,and J.Q.Zhang,“A collaborative

virtual geographic environment based on P2P and grid technologies,”Inf.Sci.,vol.177,no.21,pp.4621–4633,2007.

[25]J.Gong,J.Zhou,W.Li,and H.Lin,“Design and implementation of an

intelligent virtual geographic environment for the simulation of SARS transmission,”in Proc.ACM Int.Conf.Virtual Real.Continuum Appl.(VRCAI),2006,pp.383–386.

[26]P.M.Torrens and D.O’Sullivan,“Cellular automata and urban simu-lation:Where do we go from here?”Environ.Plann.B ,vol.28,no.2,pp.163–168,2001.

[27]J.Ernst,I.Coninx, B.Dewals,and S.Detrembleur,“Social ?ood

impacts in urban areas:Integration of detailed ?ow modelling and social analysis,”in Proc.IAHR ,2009,pp.6820–6827.

[28] E.C.Penning-Rowsell and T.Wilson,The Bene?ts of Flood and Coastal

Defence:Techniques and Data for 2003.London,U.K.:Middlesex Univ.,2003.

[29]J.Ernst et al.,“Micro-scale ?ood risk analysis based on detailed 2D

hydraulic modelling and high resolution geographic data,”Nat.Hazards ,vol.55,no.2,pp.181–209,2010.

[30]P.H.Waarts,“Method for determining loss of life caused by inundation,”

TNO,Delft,The Netherlands,Tech.Rep.B-91-1099(in Dutch),1992.[31]S.N.Jonkman,J.K.Vrijling,and A. C.W.M.Vrouwenvelder,

“Methods for the estimation of loss of life due to ?oods:A literature review and a proposal for a new method,”Nat.Hazards ,vol.46,pp.353–389,2008.

[32]J.Cirbus and M.Podhoranyi,“Cellular automata for the ?ow simulations

on the earth surface,optimization computation process,”Appl.Math.Inf.Sci.,vol.7,no.6,pp.2149–2158,

2013.

Jun Zhu received the M.S.degree in geodesy and survey engineering from Southwest Jiaotong Univer-sity,Chengdu,China,in 2003,and the Ph.D.de-gree cartography and geographic information system from Chinese Academy of Sciences,Beijing,China,in 2006.

From 2007to 2008,he was a Postdoctoral Research Fellow with the Chinese University of Hong Kong,Shatin,Hong Kong.Currently,he is an Assistant Professor with the Department of Remote Sensing and GeoSpatial Information Engineering,

Southwest Jiaotong University.His research interests include 3-D GIS and

VGE.

Lingzhi Yin received the B.S.degree in geograph-ical science from the Department of Resource and Environmental Sciences,Guangxi Teachers Educa-tion University,Nanning,China,in 2012,where she is currently pursuing the M.S.degree in cartography and geographic information system.

Her research interests include 3-D GIS and

VGE.

Jinhong Wang received the B.S.degree in survey engineering from the Department of Remote Sensing and GeoSpatial Information Engineering,Southwest Jiaotong University,Chengdu,China,in 2011,where he is currently pursuing the M.S.degree in cartog-raphy and geographic information system.

His research interests include 3-D GIS and

VGE.

Heng Zhang received the B.S.degree in survey engineering from the School of Surveying Engi-neering,Henan University of Urban Construction,Pingdingshan,China,in 2006,and the M.S.degree in geodesy and survey engineering from the De-partment of Remote Sensing and Geospatial Infor-mation Engineering,Southwest Jiaotong University,Chengdu,China,in 2010,where he is currently pursuing the Ph.D.degree in geodesy and survey engineering.

His research interests include 3-D GIS and

VGE.

Ya Hu received the M.S.degrees from the Geodesy and Survey Engineering,Southwest Jiaotong Univer-sity,Chengdu,China,in 2005.

From 2008to 2010,he was a Research Assistant with the Chinese University of Hong Kong,Shatin,Hong Kong.Currently,he is a Lecturer with the De-partment of Remote Sensing and GeoSpatial Infor-mation Engineering,Southwest Jiaotong University.His research interests include 3-D GIS and

VGE.

Zhujun Liu received the B.S.degree in geograph-ical information system from the Department of Remote Sensing and GeoSpatial Information Engi-neering,Southwest Jiaotong University,Chengdu,China,in 2013,where she is currently pursuing the M.S.degree in cartography and geographic informa-tion system.

Her research interests include 3-D GIS and VGE.