Printer Friendly

Web cellular automata: a forest fire modeling approach and prototype tool.


Rapid advances in computer processing and visualization capabilities, coupled with the decreasing costs of personal computers, have allowed a wider end-user group to access advanced GIS, spatial modeling, and visualization technologies. Consequently, the development of GIS-based applications and environmental models on standalone desktop computers and workstations within small networks has become an established practice for small businesses, consulting companies, governments, and academics. However these standalone applications and models suffer from many limitations that include platform dependence, limited end-user access, and inefficient data and information dissemination.

The use of Internet technology can overcome many limitations of stand-alone GIS applications and environmental models (A1-Sabhan et al. 2003). The Internet is an interconnected system of net works linking computers around the globe, regardless of geographic location. Internet technology has evolved rapidly during the last decade, and the Internet and world wide web (WWW) are now firmly recognized as effective means of exchanging geospatial data (Rohrer and Swing 1997; Doyle et al. 1998).

GIS applications on the Internet can be traced back to the second half of the 1990s. Internet GIS is a network-based GIS utilizing both the wired and wireless Internet to provide access to remote geospatial data and geo-processing tools (Peng 1999). The structure is based on a client/ server arrangement with functions for presentation, program logic, and database management distributed between the client and server computers. The world wide web is a networking and an information-sharing application based on the HTTP protocol that extends the Internet framework. Most Internet GIS applications use the web to exchange data, perform limited spatial analysis, and visualize results. Hence, the term Web-based GIS or Web GIS is often used for this kind of GIS extension (Peng and Tsou 2003).

Deploying GIS applications on the Internet is more advantageous than using stand-alone applications (Xie and Yapa 2006; Bellasio and Bianconi 2005; Al-Sabhan et al. 2003). Models on the Internet can be accessed by multiple users at different geographical locations, thus leading to increased involvement of stakeholders in the decision-making process. This in turn allows further communication and collaboration in decision-making (Dragicevic and Balram 2004; Mustajoki et al. 2004; Li 2006). Remote databases can be accessed in different geographical locations as well as in real time. Data security and integrity can be maintained more efficiently on data servers, by specifying varying degrees of permission for different users. Moreover, there is no need for specific GIS or modeling software on the client computer, thereby allowing application and database upgrades and debugging from a central location. By providing a consistent and user-friendly interface with essential tools and functionality for the specific application, the end-user is more focused on the tasks and goals for solving the problem. Furthermore, GIS applications and models are not dependent on a specific hardware or operating system platform; an Internet connection and web browser are the only prerequisites for accessibility.

There exists a wide range of proprietary Web GIS software that is engrained within the GIS user and developer community. Examples software include Autodesk's MapGuide (Autodesk, Inc. 2007), ESRI's ArcIMS (ESRI 2007), ER Mapper's Image Web Server (ER Mapper 2007), GE SmallWorld's Internet Application Server (GE Energy 2007), Intergraph's Geomedia Web Map (Intergraph 2007) and MapInfo's MapXtreme (MapInfo 2007). Traditional Web GIS has been mainly limited to web mapping applications that serve static maps as raster images as well as more interactive vector-based thematic maps. Specifically, the functionality of Web GIS software technology is limited to common tasks such as pan and zoom, geo-coding and buffering, spatial queries, and feature extraction. Hence, more advanced GIS analysis functionality and spatio-temporal modeling capabilities are not yet fully established as mainstream Web GIS technologies.

In parallel with the increased use of web-based GIS, the past decade has seen more active research on web-based spatial modeling and spatial decision support systems (SDSS). Increasingly, researchers are now making their spatial models accessible over the web, with examples including: oil spill modeling (Xie and Yapa 2006), simulation of industrial accidents and atmospheric dispersion (Bellasio and Bianconi 2005), land evaluation for agricultural soil protection (De la Rosa et al. 2004), exploratory data analysis of river water quality parameters (Halls 2003), management of urban soils (Hossack et al. 2004), river systems (Chang and Chang 2002), watersheds (Choi et al. 2005; Huang and Worboys 2001), hydrological land-use change impact assessment (Choi et al. 2005), flood prediction (Al-Sabhan et al. 2003), floodplain management (Stigumaran et al. 2000), livestock grazing (Mohtar et al. 2000), and soil contaminant transport (Zeng et al. 2002).

There are several challenges related to implementing models that can handle spatial and temporal components of the change process over the web. The computational load between client and server must be balanced for optimal performance. There must be sustained bidirectional communication between the client and server during the modeling process so as to update the displayed results at each time step (Huang 2003). As well, the lack of temporal GIS capabilities is an ongoing challenge facing web-based modeling efforts because the integration of Web GIS and dynamic models still needs to be adequately addressed.

Given the advantages of deploying spatial models over the web, one application area that can particularly benefit from the new web functionalities is natural hazard/disaster management and response, where both rapid information dissemination of real time data and collaborative decision making are of paramount importance (USGS 2007). Forest fire is one such natural hazard. Forest fires serve an integral role in maintaining the health and diversity of many forest ecosystems (Johnson and Miyanishi 2001), but they may also have negative socio-economic impacts linked to public health, safety, property, and resource depletion.

There are many fire models and decision support systems in existence or in continuous development (Pastor et al. 2003; Andrews and Queen 2001). Accessing these models and decision support tools over the web can provide valuable assistance to fire management agencies and stakeholders involved in making more effective management decisions in real time. Important benefits include the prevention of loss of life and property, as well as reducing the adverse effects on natural resources.

Deployments of fire-related applications over the web such as forest fire risk mapping (CFS 2007), real-time and forecast weather (Rocky Mountain Research Station 2007), hot spot detection (MODIS 2007), and burnt area mapping (NorthTree Fire International 2007) have been reported in the literature. However, the challenges associated with extending spatio-temporal fire simulation models over the web has not been adequately addressed (Eklund 2001). In this study, we argue that making such simulation capability accessible over the web (especially using a thin client web interface) can be of immense value for fire disaster management and evacuation. Fire managers and decision makers can use PDAs or handheld computers at any place with an Internet connection (e.g., fire base station) and can run simulations of the progress of fire, based on the most up to date data (e.g., weather, fuel moisture), in order to choose the most appropriate fire fighting strategy or evacuation plan for communities in danger. The objective of this study is to extend the cellular automata (CA) forest fire model over the web, thereby facilitating end-user access to complex spatio-temporal simulation modeling functionality in a flexible and user-friendly modeling environment. The forest fire behavior model integrated in the Web-CA framework was developed by Yassemi et al. (2008).

WEB GIS Background

Peng and Tsou (2003) define the technological evolution of Internet GIS as starting with static map publishing and evolving to static Web mapping, followed by interactive Web GIS and, finally, distributed GIServices. With the exception of static map publishing, these categories are all based on the same general three-tier client/server architecture consisting of client, middleware services or application servers, and web and data servers.

Static map publishing uses a two-tier architecture to distribute maps on a client Web browser as static map images. Static web mapping utilizes CGI (Common Gateway Interface) as a middleware in a three-tier architecture to link the Web browser with GIS or mapping programs on the server where a map is generated and served to a browser as an image. Interactive web mapping allows increased user interaction with the client interface and expands client-side processing capabilities and functionalities by using scripts (e.g., dynamic HTML) and/or client-side applications such as plug-ins, Java applets, or ActiveX controls. In this category, the middleware comprises CGI extensions such as Microsoft's ISAPI and ASP, Netscape's NSAPI, NeXT/Apple's WebObjects, and Java servlets or Mlaire's Coldfusion among others (Peng and Tsou 2003). Distributed GIServices use a direct communication between GIS components on the web client with other GIS components on the server without the need for a HTTP server and CGI-related middleware.

This study uses Microsoft ASP.NET (Microsoft 2007a) as the web application development tool. ASP.NET is a fully object-oriented platform built on the Microsoft .NET Framework. ASP.NET uses the Common Language Runtime (CLR) compiler shared by all .NET applications and so allows programs to be developed and authored in any programming language supported by the .NET Framework (e.g., C#, VB, C+ +,J#, Perl, Python). When an ASP.NET page is compiled, it is translated to the Microsoft Intermediate Language (MSIL) by the CLR and thereafter converted to native machine language to run on the processor. Compiled machine native code executes faster than interpreted scripting language codes.

ASP.NET Web applications are executed on a Web server configured with Microsoft Internet Information Services (IIS). ASP.NET communicates with IIS to handle requests for Web Forms pages and XML Web services. Applications are developed by using an event-driven paradigm compared to the linear top-down processing of scripting languages. Web Forms pages are rendered in the appropriate mark-up depending on the web browser detected and, hence, web applications can be developed to run on different browsers and client devices supported by ASP.NET such as cellular phones, handheld computers, and Personal Digital Assistants (PDAs). ASP.NET can also work around the stateless nature of the Web by providing facilities for saving the page and control properties as well as variables and application-specific or session-specific information between roundtrips from client to server.

Several functionalities of the .NET framework make it a suitable choice for this research study. The support of the .NET framework for different programming languages, along with support for Rapid Application Development (RAD) by the Visual integrated development environment (IDE), enables faster and easier creation of robust and scalable web models. In addition, the .NET framework supports distributed GIServices, as well as XML-based data interchange and AJAX web development technique which can make possible the development of extensions to the functionalities of the web model proposed in this study.


Cellular Automata Forest Fire Behavior Model

Forest fires can be regarded as complex systems where the pattern formation and behavior of the fire is the result of complex interaction between many factors such as landscape topography, fuel load, moisture, and local weather. Complex systems are defined by the nonlinear relationships between their interacting parts. In such systems, emergent properties are the global patterns exhibited as a result of local interactions between the individual components of the system (Manson 2001). Among the different methods to explain and understand complex systems, cellular automata (CA) models have become one of the most common approaches (White and Engelen 2000). Cellular automata are dynamic bottom-up modeling approach that can explicitly consider the spatial and temporal extent of various geographic phenomena.

Cellular automata have been extensively applied over the past two decades for modeling land-use change, urban and regional growth (Deadman et al. 1993; Yeh and Li 2001; White and Engelen 1994; Wu and Webster 2000; Clarke and Gaydos 1998; Batty and Xie 1994; and Torrens and Benenson 2005). While popular with urban and land-use change models, CA modeling has also been widely used to understand and represent a variety of complex spatial phenomena, including rainforest dynamics (Alonso and Sole 2000), plant competition (Matsinos and Troumbis 2002), forest insect propagation (Bone et al. 2006), plant invasion and dispersal (Cannas et al. 2003), habitat fragmentation (Balzter et al. 1998), landslides (Clerici and Prego 2000), and epidemic propagation and vaccination (Morley and Chang 2004), among others.

Approaches to fire simulation modeling can be divided into two general categories, with the first based on regular grid systems and the second based on continuous planes (Richards 1995). These approaches differ in their representation of the landscape and the criterion used to simulate fire behavior (Albright and Meisner 1999). Bond percolation and CA that use regular grids, and elliptical wave propagation using continuous planes, are among the most widely used approaches for simulating wildland fires. The use of cellular models to simulate fire behavior as discrete processes on a regularly spaced landscape grid has evolved into a common approach (Kourtz et al. 1977; Clarke et al 1994; Perry et al. 1999; Bendicenti et al. 2002; Berjak and Hearne 2002; De Vasconcelos et al. 2002).

Cellular automata were first introduced by von Neumann (1966) as mathematical representations of complex systems. The CA consists of a grid or lattice of cells where each cell is in one of a number of finite states. Each cell is usually represented as a discrete number. The state of a cell depends on the state of the neighboring cells and a set of transition rules. The transition rules can be deterministic, fuzzy, probabilistic, or stochastic, and they are based on the logic of the process that is being modeled. Hence, the spatial dynamics of change are influenced by these rules. The temporal component progresses in discrete time steps and the cells update their state synchronously after the transition rules are applied. There are many features of the CA that make them especially suitable for spatio-temporal modeling. One of them is the potential to generate complex patterns at a global scale from simple rules that operate at local scale. Also, they are inherently spatio-temporal, with their design structure being compatible with raster geospatial data. As such, CA designs are capable of enhancing the dynamic modeling capability of raster-based GIS (White and Engelen 2000).

A deterministic short-range CA-based fire behavior model that responds to heterogeneous environmental conditions developed by Yassemi et al. (2008) is used for this study. The fire spread rate and direction parameters, obtained from the Fire Behavior Prediction (FBP) System (Forestry Canada Fire Danger Group 1992), as well as topographic, forest fuel, and weather variables were used to calculate the fire rate spread vector for each cell. These parameters and variables are incorporated in the CA transition rules that simulate the propagation of fire between cells in a raster GIS data layer.

The developed CA fire behavior model uses a Moore neighborhood consisting of eight cells adjacent to the central cell. However, this model does not take into account expanded neighborhoods, and, as a result, it precludes fire spotting. The state of a cell at time t is defined by the proportion of the forest surface area that is burning. The cell state can range from 0 (unburned) to 1 (completely burned) along a continuous scale. The transition rules are based on the assumption that fire can spread from a neighboring cell to the central cell only when the neighbor cell is completely burning (Yassemi et al. (2008)). Once the central cell is ignited, fire travels through the cell according to the cell's rate of spread vector as depicted in Figure 1. The spread of fire from each of eight neighbors to the central cell is calculated at the end of each time step. A universal equation is developed to calculate the net burning area in a cell by accounting for contribution from all sides. Iterations for the duration of the simulation considering all cells and surrounding neighbors produce simulated fire spread across the landscape.


The CA-based fire behavior model was evaluated against a real fire event, and by comparison with fire spread simulations derived from Prometheus (Prometheus User Manual 2004), a national Canadian fire modeling tool based on elliptical wave propagation. The simulations indicated good visual and quantitative agreements between the CA model and Prometheus in complex heterogeneous conditions (Yassemi et al. (2008)).

Extension of CA Fire Model on the Web

The extension of CA models on the web has been mainly for theoretical and experimental demonstrations, rather than to solve applied spatial modeling problems. Two exceptions supporting spatial modeling with CA include WILSIM Web-based Interactive Landform Simulation Model (Luo 2007)--which is an educational modeling tool for geological landform evolution, and LEAM--Land Use Evolution and Impact Assessment Model (LEAM 2007)--which deals with dynamic simulation of land-use changes. This study presents an approach to extend the existing CA fire behavior model for distributed web access. The web CA model was developed using two methods that differ in the way raster images are created. Figure 2 presents a schematic of the architecture of the integrated environment. The first method involves the creation of a resultant image in GIF format programmatically and serving that image to the Web Client. The second method relegates the task of image creation to the GIS software. The inclusion of GIS software enhances the potential for integrating Map Servers or GIS Servers with this web application.


Image Creation in the Programming Environment A raster data layer is composed of a matrix of cells with either a quantitative or qualitative value assigned to each cell. The raster image is placed on the web server as an ASCII grid text file. The ASCII grid is read and stored in a two-dimensional array in the programming environment (automated programmatically).

In the case of modeling fire propagation, several factors such as slope, aspect, fuel load, and fire ignition are important, and their values need to be accessed in each raster cell in order to be able to implement the mathematical equations used by the Canadian Fire Behavior Prediction System (Forestry Canada Fire Danger Group 1992). Given that any raster image can be stored in a two-dimensional array, the GIS raster data layers of all the factors involved in the model can be stored in two-dimensional arrays, which in turn will provide access to the values of these factors for each individual cell (Yassemi et al. (2008)).

The programming language provides considerable flexibility in choosing the neighbors of each cell (e.g., type and size) and implementing the complex CA transition rules. After the desired number of time steps (iterations), the result is written to a bitmap file programmatically from the 2D fire array which contains all the cells on fire. The bitmap is saved as a GIF file on the web server.

Visual Studio.NET was used as the IDE for the web application development. The Web Form within the Visual Studio.NET environment in this application is made up of two components in order to separate the HTML interface from the application logic. The Web Form controls render themselves to HTML automatically when the form is displayed in the web browser. The application logic is the code-behind that interacts with the form and resides in a separate file. Due to the event-driven nature of ASP.NET, any interaction with the server-side form elements on the browser can invoke events, which in turn are captured on the client and an event message is transmitted to the server by an HTTP post where the appropriate event handler (method) is called automatically. The code-behind file (.aspx.vb) contains the CA-based fire behavior model application code in VB.NET. It includes methods for array initialization corresponding to the GIS input data layers selected by the user, code for running the simulation based on the time interval chosen, and code for creating the images to be served on the browser.

During the initialization stage, input data layers related to the fire event under study are used to initialize their respective arrays. At the end of the simulation for an indicated time interval, the updated two-dimensional array of fire is used to create the new bitmap image where the burning cells (fire value of 1) on the base layer change their color to red. The bitmap image is saved as an image file on the web server and rendered in the browser image control.

Once the user request is processed and a new image is rendered on the web page after the desired time interval, it is necessary for the user to invoke the same event again to process the simulation for the next time interval. In order to keep the simulation running, the user needs to activate a command button to cause another roundtrip after each time interval.

In order to achieve a dynamic animated visualization of the fire propagation, client-side code is used to automate the process of sending requests for each time interval. A timer is used to generate a server request by programmatically activating a command button that in turn calls the function to invoke the simulation event on the server. The timer was programmed in client-side JavaScript. As the simulation image is refreshed after the desired time interval, the timer raises an event on the client and sends another request to the server for the next time interval. This process continues until the indicated simulation end time.

Image Creation using GIS Server

The IDRISI Kilimanjaro (Clark Laboratories 2004) GIS software was used as the map server in the developed web application. IDRISI is fully COM (Component Object Model) compliant and so, by using its Application Programming Interface (API) it is possible to use any COM-compliant programming language to create custom applications and develop and integrate new modules within IDRISI.

The integration of CA modeling with the GIS software was seamlessly achieved using ASCII/raster file conversion of the input/output GIS-based files in the programming environment. Using this approach, the input data can be IDRISI raster files or ASCII grid files. The difference from the previous method (i.e., image creation programmatically) is the capacity to use a variety of GIS modules for file conversion and output map creation.

If the input data are an IDRISI raster image file, the image is converted to an ASCII grid file using IDRISI's conversion functions. This file is then read and stored in a two-dimensional array. After the desired number of iterations, the result is programmatically written to an ASCII grid text file from the two-dimensional array and then converted to a raster layer by calling the IDRISI data conversion function. The output raster layer can be converted to GIF or JPEG image files in IDRISI and streamed to the browser. Other modules of IDRISI GIS can be similarly accessed by the programming environment to provide automated access to more extensive GIS functionalities on the browser. Other GIS software with the desired functionalities can be integrated with the web model in the same manner.



Data from the 2001 Dogrib fire near Nordegg, Alberta, Canada, were used to test the Web-CA forest fire modeling prototype (Prometheus User Manual 2004). The data sets were raster-based GIS layers at 25 m spatial resolution. The fire started on September 25, 2001, and burned for 22 days, reaching a size of 9,898 ha. Figure 3a - d) presents the simulations of forest fire propagation at four different time intervals--30 min, 50 min, 70 min and 90 min, respectively. The scenario with one-point source ignition was used for the purpose of testing. Different classes of fuel types are represented on the base image behind the expanding fire polygon.

The user interface developed for the Web-CA fire behavior model was categorized into logical sections to provide users with options for data input, simulation timing, and visualization (Figure 4). An intuitive, flexible, and user-friendly interface increases the model's accessibility to a wider audience such as fire managers, researchers, and stakeholders and allows them to access GIS and modeling functionality. (1)

Data input

The model inputs are selected by the user using drop-down combo boxes. These data inputs are ASCII grid files of fuel type, slope, aspect, and the fire ignition (starting position of fire which can be point, line or polygon) raster layers. The weather data file (data stream) in text file format can also be selected by the user. Topographic, fire, and weather data can be uploaded to the server. This functionality provides the flexibility of using local and up-to-date data from different sources during simulations.

The input data requirements of the fire modeling application are raster images that can be processed as ASCII grid files. These files reside as text files in an input folder on the server computer. Hence, the alternatives of using a DBMS and external connections to databases were not necessary.

Simulation Timing

The start and end time of the simulation are selected by the user. The developed program accesses weather records corresponding to the selected period and incorporates the current hourly values for weather factors such as wind speed and direction. During the simulation, the corresponding value for each hour is used. The desired iteration length for each simulation time step is selected by the user.

In general, a smaller time step provides more realistic results as expanded neighborhoods are not considered in this model, and fire spread across more than one cell in a time step can propagate aggregation errors in the model outcome. As noted earlier, in each cell the fire spread vector (fire speed and direction) is attained by the Canadian FBP System, taking into account topography, weather, and fuel type. General, fire expert knowledge regarding weather and fire conditions can aid in determining time steps. For example, approximating a fast-moving fire event of around 30 m/min and raster cell resolution of 25 m, the fastest time for fire to travel across the cell (assuming a straight spread vector direction) would be around 50 sec. Therefore, in general, for the fire event which is slower than the approximated maximum fire speed, a time step smaller than 50 sec would be appropriate. Commonly an approximation of this variable is adequate; and in cases where simulation is done without prior knowledge of the conditions of the real fire event, choosing a generally small time step such as 15 sec would insure more accurate results.


Initialization and Simulation Run Commands

The Initialize command button invokes the events for visualization of the designated base layer and initialization of arrays for the input data layers. The Next Interval command button requests the server to begin the simulation event for the set time interval. The timer automates the process of generating server requests by programmatically activating the Next Interval button and requesting a simulation event on the server with the Click Method button.


A fixed-size image control is placed on the web form for the display of the model output. The user has the choice to view the dynamic simulations of fire spread overlaid on a desired base raster layer (e.g., fuel type or digital elevation model). The intermediate results after a specified number of iterations (i.e., animation in real time) are observed. For example, with a 10 sec time step, displaying intermediate results after 60 iterations indicates an interval of 10 min. Specifying a longer time interval between displays, results in faster processing of the simulation because there are a smaller number of requests sent to the server.


Current Conditions

The current environmental conditions are tracked by displaying information in text boxes as the simulation evolves. For example, the date and time are updated after each interval. Other variables of interest to the user during simulation--such as wind speed and direction, temperature--can be added to the display set.


The proposed Web-CA modeling approach combined an existing cellular automata forest fire behavior model with the web to provide geospatial functionalities for data processing and spatial analysis, simulation modeling, and input/output transfer interactions between clients and the web. Figure 5 shows the general framework for the Web-CA modeling approach, focusing on the prototype developed in this study and indicating the potential for linkages to 3D visualizations and real-time and multiple stakeholder involvement.

The used ASP.Net structure enabled extensive image creation and image manipulation functionalities. These allowed the raster images to be developed on the fly and served to the web browser. Hence, there was no explicit need for a dedicated Map Server for the fire modeling application. This study integrated the GIS software with the web to provide the same functionality as a Map Server in order to demonstrate the capability of the web modeling environment for GIS integration. The GIS as a back-end server would allow for access to more complex GIS functions from the web browser. However, utilizing commercial and proprietary GIS packages involves the issue of software licensing compliance. For example, Clark Laboratories (2004), as well as other commercial GIS vendors, require the clients of the GIS server to be licensed users of the software. This licensing issue has been well addressed by the Open Source technology community. Extending the GIS functionality or coupling multi-criteria evaluation procedures with the Web-CA model will justify the use of the Open Source GIS software which will overcome the software licensing limitations.

The emergence of 3D web visualization software during the past few years has provided new possibilities for seamless 3D spatial data exploration and visualization over the web (Black et al. 2007). The most prominent of these online virtual globes are Google Earth (Google 2007), World Wind (NASA 2007), Microsoft Virtual Earth (Microsoft 2007b), and, the most recent, ArcGIS Explorer (ESRI 2007). Such advanced visualization power and easy geospatial and environmental data accessibility over the web can be coupled with the developed Web-CA prototype to make it more efficient as a spatial decision tool. However, 3D visualization software is not yet suitable for complex modeling and GIS spatial analysis because there is no customized interactive user interface with essential functionalities for setting simulation parameters or uploading new input data through the user interface. In addition, 3D visualization software would constitute a thick client that can degrade the speed and model accessibility for lighter client hardware.

Research has been done in real-time data acquisition from field sensors for such applications as fire and flood modeling (Coen et al. 2006; Kremens et al. 2003; Al-Sabhan et al. 2003). However, such raw data should be processed and transformed to a usable format for model input and thereafter can be downloaded to and stored in a central web server for use in modeling applications. Real-time data update within the landscape is made possible by web mashups which retrieve data from external data sources to create new services. Some noteworthy examples of using such capability in web applications are the use of Google Earth for avian flu tracking (Butler 2006a and b), mapping disaster zones of large natural disasters such as hurricane Katrina and the Pakistan earthquake (Nourbakhsh et al. 2006), and the Regional Visualization and Monitoring System (SERVIR) which is an earth observation system providing real-time decision support for a variety of natural hazards and environmental processes (SERVIR 2007). While data fusion into a single web application--commonly called mashups--has been successful in enabling real-time data access (e.g., satellite, aerial photo, weather) and simple map overlay updates on the landscape, the more complex modeling applications require data integration and analysis before presentation of the resultant outputs. Hence, based on the input data requirements of the model, data access and processing of data into compatible formats and extents from different sources should be resolved.


This study has addressed the challenge of web-based spatio-temporal modeling by proposing a web-based cellular automata approach that extends a forest fire behavior model for use on the web. A prototype tool was developed to test the proposed approach using real forest fire data. The Web-CA approach enhances the capacity of stand-alone models by making them more accessible to experts and decision makers in synchronous and asynchronous settings for improved spatial decision support and real-time fire disaster management and evacuation. The approach provides advantages such as access to real-time geospatial and environmental data, computer platform independence, and enhanced accessibility of spatial modeling tools.

The availability and accessibility of spatial simulation modeling over the web can improve natural resources management by providing rapid information dissemination and facilitating collaborative decision making where a wider cross-section of stakeholders and decision makers can contribute to the management process. Effective real-time decisions result in rapid responses to cases of disaster and emergency, such as forest fires, extreme weather condition, flooding, and oil spills. The presented integration of spatio-temporal CA-based models with GIS and the web is a valuable prototype tool that can be integrated into a larger spatial decision support system for forest resources management.

The use of ASP.NET server-side web technology and the thin-client approach overcomes the drawbacks of the traditional CGI-related methods and provides several advantages. First, complex modeling computations are allocated to the powerful server computers. Second, the client-side has access to the fully functional GIS software on the server. Third, providing data as well as ensuring their integrity and accuracy can be managed centrally by the responsible organization. These advantages make the developed approach especially suitable for supporting various devices (ranging from desktops, laptops, PDAs, to cell-phones) with minimal requirement for fast processing speeds and computing power on the client side. Such architecture is flexible and can be modified to distribute tasks between server and client (balanced client/server), if there are changes in requirements such as client/server hardware capabilities and computing power, network traffic, and connection bandwidth.

Visual Studio.NET with Visual Basic.NET programming language was used as the integrated development environment (IDE) for developing the web application in this study. The support of the .NET's framework for the use of different programming languages, along with the support of Rapid Application Development (RAD) by Visual IDE, can be of particular benefit to Web-GIS researchers and developers by enabling faster and easier creation of robust and scalable web models. Client viewers, plug-ins, ActiveX, and custom-built controls can be easily incorporated in the ASRNET framework to provide increased client interactivity.

The developed web modeling prototype contains the essential tools and functionalities for fire modeling, enabling the end user to focus on the tasks at hand. Other modeling and GIS functionality can be provided without excessive programming effort. The modeling prototype enables uploading of local data and works with current data from multiple sources. Most important, the user controls the simulation timing and sets the time intervals for output visualization. The lack of flexible user interactive capabilities and access to specialized modeling functionality are the main drawbacks of mainstream web mapping and Web GIS software.

In its current form, the prototype is designed for a narrow audience of mainly experts in GIS and forest fire simulation or decision makers dealing with specific disaster situations. Expanding this prototype to accommodate a wider spectrum of end users, including general public, providing more realistic forest fire simulation model, and improving real-time geospatial data access and spatial analysis are all important avenues for further research.


This study was supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada Discovery Grant Program.


M-Sabhan, A., M. Mulligan, and G.A. Blackburn. 2003. A real-time hydrological model for flood prediction using GIS and the WWW. Computers, Environment and Urban Systems 27: 9-32.

Albright, D., and B.N. Meisner. 1999. Classification of fire simulation systems. USDA Forest Service Fire Management Notes 59(2): 5-12. [][Note: journal title changed to Fire Management Today in 2000.]

Alonso, D., and R.V. Sole. 2000. The DivGame simulator: A stochastic cellualr antpmata model of rainforest dynamics. Ecological Modelling 133(1-2): 131-41.

Andrews, P.L., and L.P. Queen. 2001. Fire modeling and information system technology. International Journal of Wildland Fire 10(3-4): 343-52.

Autodesk, Inc. 2007. MapGuide. [http://saarc.antodesk. com/adsk/servlet/index?siteID= 5967151 &id= 601578 0; accessed June 2007).

Balzter, H., P.W. Braun, and W. Kohler. 1998. Cellular automata models for vegetation dynamics. Ecological Modelling 107(2-3): 113-25.

Batty, M., and Y. Xie. 1994. From cells to cities. Environment and Planning. B, Planning & Design 21 : $31-8.

Bellasio, R., and R. Bianconi. 2005. On line simulation system for industrial accidents. Environmental Modelling & Software 20: 329-42.

Bendicenti, E., S. Di Gregorio, F.M. Falbo, and A. Iezzi. 2002. Simulations of forest fires by cellular automata modelling. In: G. Minati and E. Pessa (eds), Emergence in complex, cognitive, social, and biological systems. New York, New York: Kluwer Academic.

Berjak, S.G., and J.W. Hearne. 2002. An improved cellular automaton model for simulating fire in a spatially heterogeneous Savanna system. Ecological Modelling 148(2): 133-51.

Black, J., C. Arrowsmith, M. Black, and W Cartwright. 2007. Comparison of techniques for visualizing fire behavior. Transactions in GIS 11(4): 621-35.

Bone, C., S. Dragicevic, and A. Roberts. 2006. A fuzzy-constrained cellular automata model of forest insect infestations. Ecological Modelling 192(1-2): 107-25.

Butler, D. 2006a. Mashups mix data into global service. Nature 439: 6-7.

Butler, D. 2006b. Virtual globes: The web-wide world. Nature 439: 776-8.

Cannas, S.A., D.E. Marco, and S.A. Paez. 2003. Modelling biological invasions: Species traits, species interactions, and habitat heterogeneity. Mathematical Biosciences 183(1): 93-110.

CFS. 2007. Canadian Forest Service, Canadian Wildland Fire Information System, en/index_e.php, (last date accessed: July 2007).

Chang, Y.C., and N.B. Chang. 2002. The design of a web-based decision support system for the sustainable management of an urban river system. Water Science and Technology 46(6-7): 131-9.

Choi, J.Y., B.A. Engel, and R.L. Farnsworth. 2005. Web-based GIS and spatial decision support system for watershed management. Journal of Hydroinformatics 7(3): 165-74.

Clark Laboratories. 2004. IDRISI Kilimanjaro. [http://; accessed June 2007.]

Clarke, K.C., and L.J. Gaydos. 1998. Loose-coupling a cellular automaton model and GIS: Long-term urban growth prediction for San Francisco and Washington/ Baltimore. International Journal of Geographical Information Science 12(7): 699-714.

Clarke, K.C., J.A. Brass, and P.J. Riggan. 1994. A cellular-automaton model of wildfire propagation and extinction. Photogrammetric Engineering and Remote Sensing 60(11): 1355-67.

Clerici, A., and S. Prego. 2000. Simulation of the Parma River blockage by the Corniglio landslide (Northern Italy). Geomorphology 33(1-2): 1-23.

Coen, J.L., J.D. Beezley, L.S. Bennethum, C.C. Douglas, M. Kim, R. Kremens, J. Mandel, G. Qin, and A. Vodacek. 2006. A wildland fire dynamic data-driven application system. UCDHSC/CCM Report No. 238. University of Colorado at Denver and Health Sciences Center, Center for Computational Mathematics Reports. Denver, Colorado.

Deadman, P., R.D. Brown, and H.R. Gimblett. 1993. Modeling rural residential settlements. Journal of Environmental Management 37: 147-60.

De la Rosa, D., E Mayol, E. Diaz-Pereira, M. Fernandez, and D. de la Rosa, Jr. 2004. A land evaluation decision support system (MicroLEIS DSS) for agricultural soil protection with special reference to the Mediterranean region. Environmental Modelling & Software 19: 929-42.

De Vasconcelos, M.J.P., A. Goncalves, EX. Catry, J.U. Paul, and E Barros. 2002. A working prototype of a dynamic geographical information system. International Journal of Geographical Information Science 16(1): 69-91.

Doyle, S., M. Dodge, and A. Smith. 1998. The potential of web-based mapping and virtual reality technologies for modelling urban environments. Computers, Environment and Urban Systems 28(2): 137-55.

Dragicevic, S., and S. Balram. 2004. A Web GIS collaborative framework to structure and manage distributed planning processes. Journal of Geographical Systems 6(2): 133-54.

Mapper, E.R. 2007. Image Web Server. [http://www. = 26; accessed June 2007.]

ESRI. 2007. ArcIMS. [ arcgis/arcims/index.html; accessed June 2007.]

Eklund, P. 2001. A distributed spatial architecture for bush fire simulation. International Journal of Geographical Information Science 15(4): 363-78.

Forestry Canada Fire Danger Group. 1992. Development and structure of the Canadian forest fire behavior prediction system. Forestry Canada, Ottawa, ON. Inf. Rep. ST-X-3, 63 pp.

Energy, G.E. 2007. SmallWorld Internet Application Server. [ gissoftware/en/sias.htm; accessed June 2007.] Google 2007. Google Earth. [ com/; accessed July 2007.]

Halls, J.N. 2003. River run: An interactive GIS and dynamic graphing website for decision support and exploratory data analysis of water quality parameters of the lower Cape Fear river. Environmental Modelling & Software 18: 513-20.

Hossack, I., D. Robertson, P. Tucker, A. Hursthouse, and C. Fyfe. 2004. A GIS and Web-based Decision Support Tool for the Management of Urban Soils. Cybernetics and Systems 35(5-6): 499-509.

Huang, B. 2003. Web-based dynamic and interactive environmental visualization. Computers, Environment and Urban Systems 2: 623-36.

Huang, B., and M.F. Worboys. 2001. Dynamic modeling and visualization on the internet. Transactions in GIS 5(2): 131-9. Intergraph. 2007. GeoMedia WebMap. [http://www.; accessed June 2007.]

Johnson, E.A., and K. Miyanishi (eds). 2001. Forest fires: Behavior and ecological effects. San Diego, California: Academic Press.

Kourtz, P., S. Nozaki, and W. O'Regan. 1977. Forest fires in the computer A model to predict the perimeter location of a forest fire. Inf. Rep. FF-X-65. Fisheries and Environment, Canada.

Kremens, R., J. Faulring, A. Gallagher, A. Seema, and A. Vodacek. 2003. Autonomous field-deployable wildland fire sensors. International Journal of Wildland Fire 12: 237-44.

LEAM. 2007. LEAM Model. [http://www.leam.uiuc. edu/; accessed June 2007.]

Li, S. 2006. Web-based collaborative spatial decision support systems L A thecnological perspective. In: S. Balram and S. Dragicevic (eds), Collaborative GIS. Hershey, Pennsylvania, Idea Group Publishing. pp. 285-315.

Luo, W. 2007. WILSIM: Web-based Interactive Landform Simulation Model. [ landform/; accessed June 2007.]

Manson, S.M. 2001. Simplifying complexity: A review of complexity theory. Geoforum 32: 405-14.

MapInfo. 2007. MapXtreme. [http://extranet.mapinfo. com/products/overview.cfm?productid = 1849; accessed June 2007.]

Matsinos, Y.G., and A.Y. Troumbis. 2002. Modeling competition, dispersal and effects of disturbance in the dynamics of a grassland community using a cellular automaton model. Ecological Modelling 149(1-2): 71-83.

Microsoft. 2007a. ASP.NET. [ com/; accessed June 2007.]

Microsoft. 2007b. Virtual Earth. [ com/virtualearth/default.mspx; accessed July 2007.]

MODIS. 2007. MODIS Rapid Response system. [http://; accessed July 2007.]

Mohtar, R., T. Zhai, and X. Chen. 2000. A world wide web-based grazing simulation model (GRASIM). Computers and Electronics in Agriculture 29: 243-50.

Morley, P.D., and J. Chang. 2004. Critical behavior in cellular automata animal disease transmission model. International Journal of Modern Physics C Computational Physics & Physical Computation 15(1): 149-62.

Mustajoki, J., R.P. Hamalainen, and M. Marttunen. 2004. Participatory multicriteria decision analysis with Web-HIPRE: A case of lake regulation policy. Environmental Modelling & Software 19: 537-47.

NASA. 2007. World Wind, National Aeronautics and Space Administration, http://worldwind.arc.nasa. gov/, (last date accessed: July 2007).

NorthTree Fire International 2007. NorthTree Fire International,, (last date accessed: July 2007).

Nourbakhsh, I., R. Sargent, A. Wright, K. Cramer, B. McClendon, and M. Jones. 2006. Mapping disaster zones. Nature 439: 787-8.

Peng, Z.R. 1999. An assessment framework of the development strategies of Internet GIS. Environment and Planning. B, Planning & Design 26(1): 117-32.

Peng, Z.R., and M.H. Tsou. 2003. Internet GIS: Distributed geographic information services for the Internet and wireless networks. Hoboken, New Jersey: John Wiley & Sons.

Perry, G.L.W., A.D. Sparrow, and I.F. Owens. 1999. A GIS-supported model for the simulation of the spatial structure of wildland fire, Cass Basin, New Zealand.Journal of Applied Ecology 36: 502-18.

Prometheus User Manual. 2004. Prometheus User Manual, v. 3.2.3. [ software.cfm; accessed June 2007.]

Richards, G.D. 1995. A general mathematical framework for modeling two-dimensional wildland fire spread. International Journal of Wildland Fire 5(2): 63-72.

Rocky Mountain Research Station. 2007. United States Department of Agriculture--Forest Service. [http://; accessed July 2007.]

Rohrer, R.M., and E. Swing. 1997. Web-based information visualization. IEEE Computer Graphics and Applications 17(4): 52-9.

SERVIR. 2007. The Mesoamerican Regional Visualization and Monitoring System. [http://servir.; accessed July 2007.]

Stigumaran, R., C. Davis, J. Meyer, T. Prato, and C. Fulcher. 2000. Web-based decision support tool for flood plain management using high resolution DEM. Photogrammetric Engineering and Remote Sensing 66(10): 1261-5.

Torrens, P., and I. Benenson. 2005. Geographic cellular automata. International Journal of Geographical Information Science 19(4): 385 -412.

USGS. 2007. U.S. Geological Survey Natural Hazards Support System (NHSS). [; accessed July 2007.]

von Neumann, J. 1966. The theory of self-reproducing automata. Urbana, Illinois: University of Illinois Press.

White, R., and G. Engelen. 1994. Cellular dynamics and GIS: Modelling spatial complexity. Journal of Geographical Systems 1: 237-53.

White, R., and G. Engelen. 2000. High-resolution integrated modeling of the spatial dynamics of urban and regional systems. Computers, Environment and Urban Systems 24: 383-400.

Wu, F.L., and C.J. Webster. 2000. Simulating artificial cities in a GIS environment: Urban growth under alternative regulation regimes. International Journal of Geographical Information Science 14(7): 625-48.

Xie, H., and P.D. Yapa. 2006. Developing a web-based system for large-scale environmental hydraulics problems with application to oil spill modelling. Journal of Computing in Civil Engineering 20(3): 197-209.

Yassemi, S., S. Dragicevic, and M. Schmidt. 2008. Design and implementation of an integrated GIS-based cellular automata model to characterize forest fire behaviour. Ecological Modelling 210(1-2): 71-84.

Yeh, A.G.O., and X. Li. 2001. A constrained CA model for the simulation and planning of sustainable urban forms by using GIS. Environment and Planning. B, Planning &Design 28: 733-53.

Zeng, H., V.J. Alarcon, W. Kingery, H.M. Selim, and J. Zhu. 2002. A web-based simulation system for transport and retention of dissolved contaminants in soil. Computers and Electronics in Agriculture 33: 105-20.

(1) Information to access the Web-CA modeling tool can be obtained from the corresponding author.

S. Yassemi, S. Dragicevic, Spatial Analysis and Modeling Laboratory, Department of Geography, Simon Fraser University, 8888 University Drive, Burnaby, BC VSA 1S6, Canada. E-mails: <>; [corresponding author]>.
COPYRIGHT 2008 Cartography and Geographic Information Society, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2008 Gale, Cengage Learning. All rights reserved.

 Reader Opinion




Article Details
Printer friendly Cite/link Email Feedback
Author:Yassemi, Shahram; Dragicevic, Suzana
Publication:Cartography and Geographic Information Science
Geographic Code:1USA
Date:Apr 1, 2008
Previous Article:An adaptive interaction architecture for collaborative GIS.
Next Article:Architecture design and prototyping of a Web-based, synchronous collaborative 3d GIS.

Related Articles
Intricate patterns from cellular automata.
Wireless-Mobile GIS for Field-Based Researchers: Assessment and Prototype.
Computation's new leaf: plants may be calculating creatures.
What a flake: computers get the hang of ice-crystal growth.
Artificial reality and telexistence; proceedings.
Modeling and visualization for spatial decision support.
Cellular automata; a discrete view of the world.
Modelling urban development with geographical information systems and cellular automata.
The land-use evolution and impact assessment model: a comprehensive urban planning support system.

Terms of use | Copyright © 2015 Farlex, Inc. | Feedback | For webmasters