multicriteria spatial decision analysis in web gis environment

23
Multicriteria Spatial Decision Analysis in Web GIS Environment Harish Chandra Karnatak & Sameer Saran & Karamjit Bhatia & P. S. Roy Received: 8 August 2005 / Revised: 26 May 2006 / Accepted: 3 November 2006 / Published online: 1 March 2007 # Springer Science + Business Media, LLC 2007 Abstract Internet, a client/server system, is a perfect means of GIS data accessing, analyzing and transmission. The World Wide Web, FTP (file transfer protocol) and HTTP programs make it convenient to access and transfer data files across the Internet. Using Internet for GIS makes it easy access to acquire GIS data from diverse data sources in the distributed environment. The geospatial multicriteria decision analysis in a client/server environment is an important and challenging task for the GIS community because of narrow Internet bandwidth for large geospatial data sets. In the present paper, we are developing a multicriteria decision analysis tool for spatial decision making in the web GIS environment. The developed system has been demonstrated for biodiversity conservation and priorities. An attempt has been made to generate the alternative decisions based on priority vectors. The multicriteria technique of Analytic Hierarchy Process (AHP) is used to derive the eigen vectors with the given multiple constraints of conflicting criteria and aims at selecting optimal alternative from the available sets. However, the evaluation recognizes the importance of expert knowledge when assigning the weights for the best spatial priorities. Comparing within classes and alternatives using judgment and decision matrix is Geoinformatica (2007) 11:407429 DOI 10.1007/s10707-006-0014-8 H. C. Karnatak (*) Geoinformatics Division, RS & GIS Applications Area, National Remote Sensing Agency, Department of Space, Government of India, Balanagar, Hyderabad, India e-mail: [email protected] S. Saran Geoinformatics Division, Indian Institute of Remote Sensing, National Remote Sensing Agency, Department of Space, Government of India, 4-Kalidas Road, Dehradun, India e-mail: [email protected] K. Bhatia Department of Computer Science, Gurukul Kangadi University, Haridware, India e-mail: [email protected] P. S. Roy RS & GIS Applications Area, National Remote Sensing Agency, Department of Space, Government of India, Balanagar, Hyderabad, India e-mail: [email protected]

Upload: ngodang

Post on 19-Jan-2017

247 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Multicriteria Spatial Decision Analysis in Web GIS Environment

Multicriteria Spatial Decision Analysis in WebGIS Environment

Harish Chandra Karnatak & Sameer Saran &

Karamjit Bhatia & P. S. Roy

Received: 8 August 2005 /Revised: 26 May 2006 /Accepted: 3 November 2006 / Published online: 1 March 2007# Springer Science + Business Media, LLC 2007

Abstract Internet, a client/server system, is a perfect means of GIS data accessing,analyzing and transmission. The World Wide Web, FTP (file transfer protocol) and HTTPprograms make it convenient to access and transfer data files across the Internet. UsingInternet for GIS makes it easy access to acquire GIS data from diverse data sources in thedistributed environment. The geospatial multicriteria decision analysis in a client/serverenvironment is an important and challenging task for the GIS community because ofnarrow Internet bandwidth for large geospatial data sets. In the present paper, we aredeveloping a multicriteria decision analysis tool for spatial decision making in the web GISenvironment. The developed system has been demonstrated for biodiversity conservationand priorities. An attempt has been made to generate the alternative decisions based onpriority vectors. The multicriteria technique of Analytic Hierarchy Process (AHP) is used toderive the eigen vectors with the given multiple constraints of conflicting criteria and aimsat selecting optimal alternative from the available sets. However, the evaluation recognizesthe importance of expert knowledge when assigning the weights for the best spatialpriorities. Comparing within classes and alternatives using judgment and decision matrix is

Geoinformatica (2007) 11:407–429DOI 10.1007/s10707-006-0014-8

H. C. Karnatak (*)Geoinformatics Division, RS & GIS Applications Area, National Remote Sensing Agency,Department of Space, Government of India, Balanagar, Hyderabad, Indiae-mail: [email protected]

S. SaranGeoinformatics Division, Indian Institute of Remote Sensing, National Remote Sensing Agency,Department of Space, Government of India, 4-Kalidas Road, Dehradun, Indiae-mail: [email protected]

K. BhatiaDepartment of Computer Science, Gurukul Kangadi University, Haridware, Indiae-mail: [email protected]

P. S. RoyRS & GIS Applications Area, National Remote Sensing Agency,Department of Space, Government of India, Balanagar, Hyderabad, Indiae-mail: [email protected]

Page 2: Multicriteria Spatial Decision Analysis in Web GIS Environment

based on Saaty’s Pairwise Comparison. The Multicriteria Spatial Decision Support System(MC-SDSS) software development uses ASP, ArcIMS 9.0, ArcSDE9.0 and Oracle 9i dataserver in the web GIS environment. The database organization of spatial and non-spatialdata is done in the RDBMS environment using ArcSDE and Oracle 9i data server.

Keywords multicriteria spatial decision analysis . SDSS . web GIS . AHP.

biodiversity conservation

1 Introduction

The web GIS is an extension and application of client/server computing, where thegeospatial data is accessible in a shareable environment. Client/server computing describesa model for computer networking that offers an efficient way to provide information andservices to concurrent user(s) at the same time. Internet is a “connectionless” process, basedon client/server architecture [22]. In a client/server model, a client is defined as a requesterof services and a server is defined as the provider. The client/server software architecture isa versatile, message-based and modular infrastructure that is more flexible, easier to use,interoperable and scalable than centralized, mainframe, time sharing computing. In order toimplement the two-tier to n-tier architecture of client/server systems, a software applicationis separated into modular segments, and each segment is installed on hardware specific tothat subsystem. In client/server computing the system management is one of the mostimportant tasks to handle. As the applications and data are distributed across the network,however, it can be a challenge to keep configuration information up-to-date and consistentamong all the devices with secure data access.

Geospatial technology is an emerging multidisciplinary approach which involvescomputer science, geography, photogrammetry, cartography, remote sensing, surveying,GPS technology, statistics and other disciplines concerned with handling and analysingspatially referenced data. Traditional GIS can serve only dedicated users with sophisticatedsoftware and hardware with limited impact on the public. The web enabled GIS facilitatedecision making at the strategic, tactical, and operational levels; support administrativeoperations; and serves as a gateway for decision makers and general users to access thesystem conveniently and effectively. The network infrastructure and hardware specificationfor Internet GIS (based on the client/server model) provide high-speed communicationchannels for publishing and accessing geographic information through the computer network.Multicriteria decision making (MCDM) and a wide range of related analytical techniquesoffer a variety of decision making processes to expose and integrate choices with availableMCDM methods in order to solve “real-world” GIS-based planning and managementproblems. In the decision making process of real world GIS-based problems related spatialand non-spatial data, acceptable technique and are needed an interactive system toincorporate expert knowledge. Multicriteria spatial decision support systems are part of abroader field of spatial decision support systems (SDSS). Several application specificframeworks for designing MCSDSS have been proposed [3]–[5], [8], [12], [19]. TheMCSDSS in a sharable framework can solve the real world spatial decision problem mostefficiently.

The increasing exploitation of forest and natural resources are of extreme concern toecologists in India. Geospatial techniques are important tools to cater to the need ofdecision makers in the area of eco-development and forest management. These tools canalso help study and to characterize the biodiversity of particular regions. Concern over the

408 Geoinformatica (2007) 11:407–429

Page 3: Multicriteria Spatial Decision Analysis in Web GIS Environment

role of human activity on our environment has increased the demand for integrated,spatially-distributed, environmental models that address the interactions of human activity,the terrestrial biosphere, climate and forests. Furthermore, the widespread availability ofgeographical information systems (GIS) supports spatial data processing and analysis toincrease the accessibility of spatial models. As a result, there has never been a greater needfor decision support tools to help in evaluating the applicability of complex environmentalmodels.

In the present study we are demonstrating the technological solution used in thedevelopment of the multicriteria spatial decision support system in the web GISenvironment for Biodiversity conservation and prioritization. For the application develop-ment and demonstration we have taken data sets of the Department of Space (DOS) and theDepartment of Biotechnology (DBT), Government of India, collaborative project onBiodiversity characters and landscape level using remote sensing and GIS. During theDOS-DBT project spatial data on forest/vegetation and land use are generated usingsatellite remote sensing data (IRS 1D LISS III) through digital classification. The spatialand non-spatial data from other ancillary data sources are combined to generate habitatmaps. Landscape analysis for determining the parameters like fragmentation, porosity,proximity and other patch characteristics have been used to derive disturbance index usingproximity from settlement and roads. The knowledge base (as available in literature) withrespect to ecosystem uniqueness, species richness, and biodiversity value is used to createattribute information of the composite strata of vegetation type and disturbance regimes.The terrain complexity and disturbance index were spatially combined with the aboveknowledge base to model the biological conservation prioritization areas.

2 Approach

2.1 Multicriteria spatial decision analysis

Spatial decision analysis is a set of systematic procedures for analysing complex spatialdecision problems. The strategy of decision analysis is to divide the original problem intosmall parts, analyse each part and integrate them logically to produce a meaningful solution[10]. The decision making process itself is a broadly defined term with importance in manyfields, such as social, economic and natural resource management and disaster managementincluding GIS. Spatial decision analysis is a specific subclass of the decision analysisprocess where the decision maker has to choose the best alternative from sets ofgeographically defined alternatives (events), on the basis of multiple, conflicting andincommensurate evaluation criteria. In the geographically defined alternatives the finaldecision also depends upon the spatial arrangement of alternatives (spatial variability). Mostof the real world spatial problems give rise to Geographic Information System (GIS) basedMulticriteria Decision-Making (MCDM) or Multicriteria Decision Analysis (MCDA) [10].The evaluation and ranking of alternatives by MCDM techniques is based on associatedcriteria values, objectives and preferences of the different decision makers. Spatialmulticriteria analysis is vastly different from conventional MCDM techniques because ofits additional explicit geographic component. In comparison with conventional MCDManalysis, spatial multicriteria analysis requires information on criterion values and thegeographical distribution of alternatives in addition to the decision maker’s preferences in aset of evaluation criteria. In the spatial multicriteria decision analysis two concerns are of

Geoinformatica (2007) 11:407–429 409

Page 4: Multicriteria Spatial Decision Analysis in Web GIS Environment

vital importance: (1) the GIS component (e.g., data acquisition, storage, retrieval,manipulation, and analysis capability); and (2) the MCDM analysis component (e.g.,aggregation of spatial data and decision makers’ preferences into discrete decisionalternatives) [3], [6].

Simon [21] presents one decision flow chart for multicriteria decision making (Figure 1);showing three phases of decision making, i.e., intelligence phase, design phase and choicephase. Data acquisitions, processing and examining are done in the intelligence phase;formal modeling/GIS interaction is the design phase to develop a solution set of spatialdecision alternatives. The integration of decision analytical techniques and GIS functionssupports the design phase significantly. The choice phase involves selection of theparticular alternative from available set. In this phase, specific decision rules are used toevaluate and rank the alternatives. The three stages of decision making do not necessarilyfollow a linear path from intelligence to design, then to choice [10].

2.2 AHP for multicriteria spatial decision analysis

Thomas Saaty of the University of Pittsburgh has developed the Analytic Hierarchy Process(AHP) based on three principles: decomposition, comparative judgment and synthesis ofpriorities [17], [18]. AHP is a mathematical method of analyzing complex decisions prob-lemwith multiple criteria. The decomposition principle of AHP requires the decision problemto be decomposed into hierarchy that captures the essential element of the decision problem.The comparative judgment principle of AHP requires pair-wise comparison of the de-composed elements within a given level of hierarchal structure with respect to the next higherlevel. The synthesis principle of AHP takes each of the derived ratio scale local priorities inthe various levels of hierarchy and constructs a composite set of priorities for the elements atthe lowest level of the hierarchy. The standard method used to calculate the values for theweights from an analytic hierarchy process (AHP) matrix is to take the eigenvector corres-ponding to the largest eigenvalues of the matrix, and then to normalize the sum of thecomponents equal to one [17], [18]. The detail description and implementation of AHP forspatial decision making is explained in coming sections.

The Analytical Hierarchy Process follows the nine phases for multi-criteria decisionanalysis. The graphical representation of which is shown in Figure 2. In the next section we

Problem Definition

ConstraintsEvaluation criteria

Decision Matrix

Decision Rules

Sensitive analysis

Reconditions

Alternatives

Decision Makers preferences

Cho

ice

phas

e D

esig

n ph

ase

Inte

llige

nce

pha

se

GIS

M

CD

MM

CD

M\G

IS

Figure 1 Decision flowchart forspatial multicriteria analysis [10].

410 Geoinformatica (2007) 11:407–429

Page 5: Multicriteria Spatial Decision Analysis in Web GIS Environment

give a brief discussion about each phase of the Analytical Hierarchy Process (AHP) followsin this study.

1. In any decision making process the decision maker has to identify the goal or ultimateobjective of decision problem. Similarly in Analytical Hierarchy Process, first thedecision maker’s has to identify the objective and decision problem. For example—inour case we have to identify and select the best site (spatial location) for biodiversityconservation and priorities in study area.

2. Simon [21] presents a model of the decision making process (Intelligence, identifyingthe problem, Design, generating and evaluating alternatives, and Choice, selectingalternatives). The AHP supports the choice phase. The main objective of this phase isto select one particular alternative from a set of known options. Therefore, the first stepin the analytical hierarchal process is to list all the alternatives.

3. After that we review the threshold levels of all the alternatives. Any alternatives notmeeting threshold levels are dismissed.

Figure 2 Nine phases of AHP.

Geoinformatica (2007) 11:407–429 411

Page 6: Multicriteria Spatial Decision Analysis in Web GIS Environment

4. In the third phase of AHP we define the criteria that will be used to judge the alternatives.5. After defining alternatives and criteria we develop a decision hierarchy. This hierarchy

consists of at least three levels, i.e., a goal, criterion and alternatives. The hierarchyrepresents the structure of the decision problem and forms the basis of the comparisons.

6. The next step in the AHP process is pair-wise comparison alternatives, i.e., for eachcriterion; the decision maker compares all the alternatives pair-wise. The decisionmaker can make numerical or verbal judgments. In the verbal mode, statements areselected varying from ‘equally preferred’ to ‘greatly preferred’ [18]. In the numericalmethod, the decision maker selects a score on a scale of one to nine (Table 1).

For example, the score three indicate that one alternative is three times as preferable tothe other alternative. In the present study the numerical method is used.7. The next step in AHP is to determine the relative importance of each criterion. The

decision maker compares all criteria pair-wise and decides which criterion is moreimportant, and to what extent. Based on these judgments, the importance of eachcriterion is calculated by applying similar computation rules as used in calculating therelative priorities of the alternatives.

8. Now the overall priorities of the alternatives are determined by means of a linearadditive function, in which the relative priorities for an alternative are multiplied by theimportance of the corresponding criteria and summed over all criteria.

9. The last step of AHP is sensitivity analysis. Showing the robustness of the overallpriority rating. Sensitivity analysis shows, for example, to what extent the overallpriorities are sensitive to changes in the importance of criteria. The more stable theranking of the alternatives, the more confident the decision maker will be about theproposed choice.

When the AHP method is applied for solving spatial multi-criteria decision problems iscalled Spatial-AHP. In the present study we have implemented AHP techniques in the webGIS environment for biodiversity conservation and prioritization.

2.3 Spatial decision analysis in web environment—case study

Traditional GIS can serve only dedicated users with sophisticated software and hardwarecausing limited impact on the public. Internet technology as a digital communicationmedium enhances the capability of GIS data and software application by making them more

Judgment scale for pair-wise comparisons

Description ScaleEqually preferred 1Equally to moderately 2Moderately preferred 3Moderately to strongly 4Strongly preferred 5Strongly to very strongly 6Very strongly preferred 7Very strongly to extremely 8Extremely preferred 9

Table 1 Judgment scale for pair-wise comparisons [18].

412 Geoinformatica (2007) 11:407–429

Page 7: Multicriteria Spatial Decision Analysis in Web GIS Environment

accessible and reachable to a wider range of users, planners and decision makers. The webenabled GIS facilitates decision making at the strategic, tactical, and operational levels;supports performance of administrative operations and serves as a gateway for decisionmakers and general users to access the system effectively.

Geospatial analysis such as multicriteria decision analysis in a web environment iscomplex and challenging task because of complex data structure and lack of largebandwidth on the Internet. We have developed one Multicriteria Spatial Decision SupportSystem for biodiversity conservation and priorites using the AHP technique, which can beextended for other decision analysis techniques such as SCP, and Goal programming etc.

The overall objective of developed SDSS is to select a best site for biodiversityconservation and prioritization. Biodiversity conservation priorities are one of the complexissues for conversation authorities. Various ecological and socio-economic drivers governthe spatial distribution of biologically rich communities. These drivers are an importantinput to the modelling process with criteria of different ranks and probable weights in orderto arrive at a decision making process for the conservationist and planners. Any decisionmaking process starts with the identification of a decision problem. In the present study ourobjective is multicriteria decision analysis for biodiversity conservation priorities. In theavailable literature it is shown that different landscape models are developed to characterizebiodiversity, which addresses only structured problems on biodiversity. But until todaythere is no means to address the solution for semi-structured problems having variousdegrees of uncertainty on biodiversity, i.e., where computer based models can interactdirectly with the biodiversity experts to generate a knowledge base for biodiversityconservation priorities. The measurement of biological rich areas is dependent on a decisionmaker’s preferences and the available spatial and non-spatial data.

The study is based on the landscape model developed by [16] for biodiversitycharacterization at landscape level. The data set used for the study is taken from the outputof the Department of Space (DOS) and the Department of Biotechnology (DBT); acollaborative project entitled “Biodiversity characterization at landscape level using RemoteSensing and GIS.” The data sets of the Nokrek Biosphere reserve forest have been taken forstudy purposes. In the next section we describe the implementation of AHP and thelandscape model for multicriteria decision analysis in web the GIS environment.

2.3.1 Decision alternatives/factors

The main objective of this phase is to select one particular alternative from a set of knownoptions. In real world spatial decision problems the decision maker has to select the bestgeographical location in the area of interest. The geographical location in GIS data format isrepresented by line, point or polygons (in vector data models) and by pixel or set of pixels(in raster data model). Therefore, in the spatial decision problem, listing the alternatives isquite difficult because the numbers of alternatives are very large. Here we are taking thevegetation type map as an alternative layer and each vegetation type is represented as one ofthe factors. Here selecting the best factor does not means that the selected vegetation typehas the highest biological richness, but it means that the selected vegetation type is inhighest priority for deciding the most biologically rich areas. The following vegetationtypes are available in the study area:

& Bamboo—BAMB& Moist mixed deciduous—MMD& Tropical evergreen—TEVG

Geoinformatica (2007) 11:407–429 413

Page 8: Multicriteria Spatial Decision Analysis in Web GIS Environment

& Tropical semi evergreen—TSEV& Sub tropical evergreen—STEV

2.3.2 Decision criteria

In spatial multicriteria decision analysis the evaluation criteria are associated withgeographical entities and the relationship between them; therefore, the evaluation criteriacan be represented in the form of maps or GIS layers. The criteria map can again becategorized in two types, i.e., evaluation criteria maps and constraint maps. According to[10] the evaluation criteria maps are the GIS maps that contain unique geographicalattributes of the alternative decisions that can be used to evaluate the performance of thealternatives. A constraint map displays the limitations on the value that attributes anddecision variables may assume. The evaluation criteria maps are also referred to as attributemaps, data layers and thematic maps, etc. in GIS terminology.

There are certain parameters, which determine spatial biodiversity distribution. It isessential, therefore, for decision makers to vary these parameters as per the localknowledge. In the landscape model developed by [16] the required spatial data for derivingthe biologically rich areas is already standardized. The following spatial layers have beentaken as criteria maps in the present study:

& Ecosystem uniqueness—EU& Terrain complexity—TC& Species richness—SR& Disturbance Index—DI& Biodiversity value—BV

2.3.3 Decision hierarchy

After defining alternatives and criteria for decision making, the system developer teaches adecision hierarchy. This hierarchy consists of at least three levels, i.e., a goal, criteria andalternatives. These elements are represented in a tree structure. The hierarchy represents thestructure of the decision problem and forms the basis of the comparisons as described inFigure 3.

Figure 3 Decision hierarchy.

414 Geoinformatica (2007) 11:407–429

Page 9: Multicriteria Spatial Decision Analysis in Web GIS Environment

2.3.4 Implementation of AHP in web GIS environment

Geospatial analysis such as spatial multicriteria decision analysis in the Internet GISdomain is the concern of the latest research and development in GIS and informationtechnology known as Geoinformatics. The spatial decision support systems in the webenvironment provide an interactive system to the decision maker for group decision makingon a sharable platform. This study demonstrates the real implementation of MCDA fornatural resource management in the web environment. The data used in the present studyare of national importance and restricted under the Survey of India map policy. To accessthe system there is authentication by user name and password. This multicriteria spatialdecision support system (named BioconsSDSS) is demonstrated for small areas, i.e., NokrekBiosphere reserve forest of Meghalaya State of Northeast India, but can be implemented forother areas.

In the developed SDSS, the database generated during the DOS-DBT project is stored ina database server in the RDBMS environment. The best site for biodiversity conservationand priorities is derived using five criteria (discussed in Section 2.3.2). The generation ofcriteria map is based on the landscape model given by [16] where the biological richness atlandscape level is determined as a function of ecosystem uniqueness, species diversity,biodiversity value, terrain complexity (TC) and disturbance index (DI). The mainparameters such as EU, SR and Biotic disturbance (BD) come from ground-basedobservation. Various vegetation types available in the terrain are evaluated, and it is alsodependent on local information. The landscape model finally calculates the BiologicalRichness (BR) as below:

Biological Richness (BR)= ∫ {Ecosystem uniqueness, species richness or diversity,biodiversity value, terrain complexity and disturbance index}

BR ¼Xn

i¼1

DIi �Wti1 þ TCi �Wti2 þ SRI �Wti3 þ BVi �Wti4 þ EUi �Wti5ð Þ

Where DI=Disturbance Index, BR=Biological Richness, TC=Terrain Complexity SR=Species Richness, BV=Biological Values, EU=Ecosystem Uniqueness, Wt=Weight.

In this model the biological richness map is generated by a simple weighted addition ofthe criteria. The weightings wt is fully dependent on conflicting choices and fieldknowledge of the decision maker. Here often the weighting method is not standardized andmany times it can give conflicting results. For calculating relative importance of the criteriawe are introducing the AHP pair-wise comparison method for group decision making in theweb environment. The decision maker can do value addition on decision criteria andalternatives by assigning weight wt, based on his/her local field knowledge, using Saaty’spair-wise comparison table. The derived criteria and alternative maps are available atremote servers along with attributed data of sample plot information. For organizing spatialand non-spatial data we have standardized and designed a database structure. The spatialdata is organized by using Geodatabase and stored in Oracle9i data server.

For deriving the final output map we add one SDSS shell in between the final output anddata servers. The SDSS shell is basically a software application developed by using ESRIArc Internet Map server and Active Server Pages (ASP) and JAVA programming language.The SDSS shell developed by using ASP provides a lighter version and does not requireany plug-in software at the client end. The SDSS shells calculate overall priorities ofalternatives using AHP, and send it to a spatial shell as input for deriving BR map. All theprocessing is done on the real-time basis so that concurrent users can do the same analysis

Geoinformatica (2007) 11:407–429 415

Page 10: Multicriteria Spatial Decision Analysis in Web GIS Environment

with the same data sets. The final output is fully dependent on the decision maker’s choice,preferences and field knowledge.

2.3.4.1 System architecture and implementation BioconsSDSS is based on state-of-artclient/server computing technology. The overall flow diagram of the computingenvironment is shown at Figure 4. The system is based on multi-tiered architecture whichcan be divided into two broad categories, i.e., client end process and server end process.The server end processes can again divide into two parts—(1) Application server and, (2)Data server. BioconsSDSS is an Internet GIS based application where most of the GISfunctions are available on the web browser. The GUI of BioconsSDSS provides the facilityof selecting models and areas of interest, and the system takes the decision maker’s inputfor decision analysis. After pair-wise comparison of decision alternatives and criteria, thesystem calculates the overall priorities of the alternatives and this priority goes as input inthe GIS environment. The application server processes the input with available data sets indata server and sends it to map server; map server launches this output for the web browserby using the available map service in the service registry of the map server. The JAVAServlet engine is used for connectivity between a map server and a web server. In an ASPbased application most of the programming is done using the server side programminglanguage. Multicriteria decision models like AHP are implemented by using VB script andJAVA for vector and raster data sets respectively. The geospatial data is organized asGeodatabase in the RDBMS environment which is one of the best solutions for distributedGIS applications. In the decision making process the SDSS shell prepares one SDSS layerat the back by overlay operation for vector data; in each operation it generates a large

Web interface

Model Selection Input for multicriteria decision analysis

Output generation

Web Server (IIS)

Map server (ArcIMS)

Application Server

Model Management

AHP & SCP

Applicationdevelopment

ASP & JAVA

LandscapeModel (BCLL)

Out side BioconsSDSS

Non-spatial Data(Oracle 9i)

Spatial Data(Geodatabase in

Oracle 9i)

Spatial file server

ArcSDE

Internet GIS environment

Client

Server

Database Server

Firewall protection

SDSS Shell

Spatial server

JAVA servlets engine

Figure 4 System architecture of BioconsSDSS.

416 Geoinformatica (2007) 11:407–429

Page 11: Multicriteria Spatial Decision Analysis in Web GIS Environment

number of records in Geodatabase. The complex geospatial analyses such as MCDA, forthis type of data set takes more time in geoprocessing, therefore, the performance andtuning of the GIS data can be improved by using the RDBMS function. The BioconsSDSSgives its output in two ways, i.e., one in non-GIS environment and another in GISenvironment. The statistical report and non-GIS output are in simple HTML file format.GIS environment of BioconsSDSS provides GIS tools for geospatial analysis and queryingsuch as panning, zooming, query builder, layer addition and identify tool.

The database server of BioconsSDSS contains the spatial and non-spatial data of Indianbiodiversity, which is collected from diverse data sources in different data format. Therewas a need to standardize the data structure and format for MCDA and the development ofa related Information System. The Online Analytical Processing (OLAP) and relatedconcepts of RDBMS are used for better performance and security of data. A conceptualarchitecture of a data server is given at Figure 5.

3 Results and discussion

Internet GIS is a means for GIS users to exchange GIS data, conduct GIS analysis andpresent GIS output in the form of maps. The Internet GIS application may provide simpleGIS visualization environment, spatial query tools or geospatial analysis functionality.Multicriteria spatial decision analysis in Internet GIS environment is a specific type ofgeospatial analysis where interaction between decision maker (at client end) and GIS dataand application (at server end) is required. Spatial Decision Support (SDSS) usingmulticriteria spatial decision analysis is commonly considered an application-specificsoftware solution. The MC-SDSS is used in solving complex spatial problems wherealternative decisions require consideration. Wellar [24] and Crossland et al. [25] showedthat the use of GIS as a type of SDSS reduced the decision time and increased the accuracyof individual decision makers, while [13] emphasize that the Internet provides an idealplatform for non-experts to realize the power and benefits of GIS. Integrating thesetechnologies in a web GIS-based SDSS has the potential to increase the use andaccessibility of spatial data, as well as the accuracy and efficiency of decision making.

The integration of MC-SDSS in the Internet environment is the emerging area. TheSpatial Decision Support System in the Internet domain uses Internet protocols, GISfunction and multicriteria technique for spatial data analysis and visualization for spatialdecision making. A common motivation for making SDSS accessible on web is to supportgroup decision making. Public participation through Web based SDSS is promoted by [9],[23] and [20]. Rinner [14] categorized web based SDSS in three major categories: Server-side Web-SDSS, Mixed client and server side Web SDSS, and Client side Web SDSS.

In Server side Web SDSS all the decision analysis related operations are executed at theserver end only. The client can visualize the output only in simple web browser. Somepopular examples of Server side Web SDSS are: Open spatial decision making on theInternet (OSDM) [3]; Multi criteria geographic information system (MC-GIS) [11];Recycling decision support system [2].

Web based spatial decision support for a mixture of client and server side Web SDSStypically supports group decision making. Compared with strictly server side WebSDSS,implementing through Java applets allows clients more interactively. The major develop-ments in this category are Virtual Slaithwaite [9]; VegMan and JavaAHP [23]; andCollaborative WebSDSS for land use change assessment [20].

Geoinformatica (2007) 11:407–429 417

Page 12: Multicriteria Spatial Decision Analysis in Web GIS Environment

The third category is strictly client side Web SDSS that represents more recent devel-opments in web based SDSS. The client side web SDSS suggests advanced visualizationand multi-criteria evaluation methods, and integrates Web SDSS with state-of-the-art inother geographic information techniques [14]. Jankowski et al. [7] combine cartographicvisualization techniques with multi criteria decision analysis and data mining; Andrienkoand Andrienko [1] propose intelligent user guidance in interactive decision support; andRinner and Malczewski [15] introduce decision strategies in Web SDSS.

The developed web SDSS in present study is based on the server side and also a mix ofclient and server side technologies. The spatial data is available at a centralized server in astandard data format. The database has been generated during the national level DOS-DBTcollaborative programme on biodiversity characterization at landscape level. Biodiversityconservation priorities are one of the complex issues for conservation authorities. Variousecological and socio-economic drivers govern the spatial distribution of biologically richcommunities. The developed system (BioconsSDSS) addresses the semi-structured problemwith various degrees of uncertainty on biodiversity, i.e., where computer based models caninteract directly with the biodiversity experts to generate a knowledge base for learningbiodiversity conservation priorities. The developed web enabled SDSS provides a GISenvironment in a simple web browser at the client end; the decision maker’s interactionwith available data sets; and standard multicriteria analysis technique, i.e., AHP.BioconsSDSS has two major parts in application at server level: first it takes input fromthe decision maker and calculates overall priority of the alternatives using AHP, then theoutput from AHP used as input for geo-processing and spatial decision analysis.

Figure 5 Conceptual architecture of database server.

418 Geoinformatica (2007) 11:407–429

Page 13: Multicriteria Spatial Decision Analysis in Web GIS Environment

Some of the screen shot of the developed system are shown in Figures 6, 7, 8, 9, 10, 11,12, 13, 14, 15 and 16. The developed SDSS provides a facility to select factors and criteriawith initial weight based on decision maker’s field knowledge (Figure 6). Once the decisionmaker provides this input to the system the application opens a new window for generatinga decision matrix (Figure 7). The decision maker can select particular factor/alternative forpair-wise comparison corresponding to each criterion using Saaty’s pair-wise comparisontable (Figure 8). For each alternative the system calculates relative priorities using AHP(Figure 9). After pair wise comparison of alternatives the application generates a decisionmatrix (Figure 10). After generating decision matrix for alternatives the application opens anew window for pair-wise comparison of decision criteria and calculates the relativeimportance of the criteria’s (Figure 11). The overall priority of alternatives/factors usingAHP is shown in Figure 12. Based on this multi-criteria decision analysis, the SDSS Shellruns the model in web browser itself and display SDSS output in web GIS environment(Figure 13, Figure 14). In the Figure 13 the vegetation type map is shown where inFigure 14 the SDSS out map is displayed. The SDSS output map is showing the prioritiesof the area from least priority to greatest priority. For example the red color in Figure 14shows the highly biological rich sites where yellow color shows least priority areas forbiodiversity conservation. The main area shown in red color is core boundary of reserveforest. The developed SDSS also provides all or almost all GIS functionality in simple webbrowser for spatial querying and analysis like query builder and selection of particular areas(Figure 15) and calculation of statistical report (Figure 16) etc.

The related development by [23] describes a Web-based information and decisionsupport system, VegMan, for regional vegetation management. The system was designed tosustain biodiversity and native vegetation in Queensland, Australia. The Internet is used todisseminate information and provide access to analytical tools. VegMan is a client/serversystem using HTML pages and Java applets as its user interface. On the server side,

Figure 6 Selecting factors and criteria with initial weight.

Geoinformatica (2007) 11:407–429 419

Page 14: Multicriteria Spatial Decision Analysis in Web GIS Environment

Figure 7 Decision maker’s input for MCDA.

Figure 8 Window for user’s input (pairwise comparison of factors).

420 Geoinformatica (2007) 11:407–429

Page 15: Multicriteria Spatial Decision Analysis in Web GIS Environment

Figure 10 Generated decision matrix.

Figure 9 Calculation of relative priorities (analysis).

Geoinformatica (2007) 11:407–429 421

Page 16: Multicriteria Spatial Decision Analysis in Web GIS Environment

Figure 12 SDSS Shell: windows showing final AHP output.

Figure 11 Calculation of relative importance and decision matrix for criteria.

422 Geoinformatica (2007) 11:407–429

Page 17: Multicriteria Spatial Decision Analysis in Web GIS Environment

VegMan consists of three software components, all of which are implemented in Java. Arule-based system developed with JESS answers queries for available vegetationmanagement strategies, assembling HTML documents to return to the user. WebMap isan Internet mapping toolkit used to provide maps with the text documentation byintegrating the WebMap applet into the HTML documents. The third component isJavaAHP, an implementation of the analytical hierarchy process (AHP) multi criteriaevaluation method.

In the present study the developed Web-SDSS is an integration of the landscape modeldeveloped by [16] and multicriteria decision analysis technique AHP [18] to address thesemi-structured problem with various degrees of uncertainty on biodiversity. The client sideenvironment is based on ASP and JAVA where, in ASP based application the simple webbrowse is sufficient to access the system, while in JAVA base application the client requiresJAVA plug-in at a local machine.

A web enabled multi criteria spatial decision support system for sorting biodiversityconservation priorities have been developed and tested for different data sets of Indianbiodiversity. The main advantage of this system is that it integrates the landscape modelwith AHP for deriving biologically rich sites in a group decision making environment. Thevegetation type map, species richness map, fragmentation map, disturbance index andbiodiversity value map are derived by using the landscape model, and stored asGeodatabase in a central repository along with the sample plot data. The decision makercan interact with these spatial layers and, after value addition; the landscape modelgenerates a new BR map based on the decision maker’s choice and preferences.

At the database server level the Geodatabase architecture is adopted with OLAP usingOracle 9i. The designed database architecture and standards are not specific for this application.Other related developments using this database are Biodiversity Information System http://www.bisindia.org) and Plant species Information System http://www.bisindia.org/phytosis).

Figure 13 SDSS Shell-vegetation type map with legend.

Geoinformatica (2007) 11:407–429 423

Page 18: Multicriteria Spatial Decision Analysis in Web GIS Environment

The developed MC-SDSS is available in the Internet GIS domain under the web portalof Indian Biodiversity Information System (http://www.bisindia.org). An attempt has beenmade to develop one generic SDSS in web environment where the system can define theproblem and give the facility to calculate the relative importance of decision alternativesand criteria.

The multicriteria spatial decision support system in the Internet GIS domain generallygives a slow performance because of narrow bandwidth and large size of geospatial data.We have tested BioconsSDSS for a larger area also, such as whole Northeast India andAndaman Islands and it was observed that the system works very slowly for the larger area.

Internet GIS has arrived yet is constantly evolving technology. The fully interoperableInternet GIS becomes more promising as Internet standards and technologies rapidly grow.

The advent of Java-a portable, object oriented Internet language promises to removemany of the constraints inherent in early www protocols and further extends the capabilitiesof web-based data browsing systems. By moving much of the requisite display, processing,and analysis functional to the client end of the Internet connection, performance delayscaused by server overload and Internet bandwidth limitations will be greatly reduced.

The new concept of Geodatabase is pioneering towards fully interoperable Internet GIS. Inthe Geodatabase the spatial data is organized and stored in a relational database managementsystem (RDBMS), where the available non-spatial data in RDBMS can easily link withspatial data and all the functions and capability of a database management system can be fullyused for GIS data. The spatial or non-spatial data stored in any database management systemalways perform better and secure data access in a multi-user environment.

The success of Spatial Decision Support (SDSS) system in Internet GIS domain is fullydependent on the performance of the application. A large number of multicriteria SDSSsolutions are available commercially or freeware for desktop applications. Development of

Figure 14 SDSS map (derived) with legend.

424 Geoinformatica (2007) 11:407–429

Page 19: Multicriteria Spatial Decision Analysis in Web GIS Environment

Figure 15 Running the sample query and showing selected area.

Figure 16 Window for statistical report.

Geoinformatica (2007) 11:407–429 425

Page 20: Multicriteria Spatial Decision Analysis in Web GIS Environment

SDSS in web GIS environment with full spatial analysis capability is still evolving. Theimplementation of online analytical processing (OLAP) and related techniques at spatialdata server level will certainly improve the performance and analytical capability of webbased spatial decision support.

In a web environment, performance is usually the most important factor, thus adeveloper should keep in mind the network performance when designing the database. Thedatabase normalization and indexing performs best for an Internet GIS application.Threading and session management are also two most important considerations that affectperformance and scales.

The recent development in web GIS using Asynchronous Java script and XML (AJAX)model is a new revolution for the GIS community. AJAX techniques greatly increase theperformance and visualization power of web GIS application. Implementation of AJAX forGIS visualization in web based SDSS will certainly improve the performance and successof the system.

4 Conclusion

This paper demonstrates the web base spatial decision support system using multicriteriaspatial decision analysis for biodiversity conservation and priorities. Web based SDSS i.e.,BioconsSDSS, is developed by integrating standard multicriteria decision analysis techniqueAHP with a landscape model for biodiversity characterization. An attempt has been made togenerate the alternative decisions based on priority vectors. The multicriteria technique ofAnalytic Hierarchy Process (AHP) is used to derive the eigen vectors with given multipleconstraints, conflicting criteria and aims at selecting the optimum from an available set ofalternatives. However, the evaluation recognizes the importance of expert knowledge whileassigning the weights for the best spatial priorities. The developed web based SDSS isaccessible to multi-user environments for group decision making. Web based technologies likeASP, JAVA and internet map server ArcIMS are used to develop the application. In the serverside application most of the geoprocessing is done at the server end which makes slowly forlarger area while the mix of the client and server base architecture gives the web based SDSSbetter performance. The spatial data with attribute data are organized into the RDBMSenvironment as a central repository. A database structure and standard for Indian biodiversitydatabase is designed, which can be used for the development of SDSS or related informationsystems. The developed web based SDSS, i.e., BioconsSDSS, provides an interactive GISenvironment in thin as well as thick client for basic GIS operations and querying.

The main limitation of the system is slow performance in larger areas, which can beimproved by using new technologies such as AJAX for GIS visualization and JAVA baseclient application development. Due to the data security policy, it is not possible todownload the original spatial data at the client end but, interactive GIS visualization in theweb browser is available with the most important GIS tools. A part of application and datacan be downloaded as static maps.

Acknowledgements The authors duly acknowledge the DOS-DBT project team for providing the base maplayers used in this study. Special thanks to Dr. M. C. Porwal, Scientist, IIRS, Deharadun India, for theproviding necessary data on the Nokrek biosphere reserve forest.

426 Geoinformatica (2007) 11:407–429

Page 21: Multicriteria Spatial Decision Analysis in Web GIS Environment

References

1. N.V. Andrienko and G.L. Andrienko. “Intelligent support for geographic data analysis and decisionmaking in the web,” Journal of Geographic Information and Decision Analysis, Vol. 5(2):115–128,2001.

2. H.K. Bhargava and C.G. Tettelbach. “A web-based DSS for waste disposal and recycling,” Computers,Environment and Urban Systems, Vol. 21(1):47–65, 1997.

3. S.J. Carver. “Integrating multi-criteria evaluation with geographical information systems,” InternationalJournal of Geographical Information Systems, Vol. 5:321–339, 1991.

4. J.T. Diamond and J.R. Wright. “Design of an integrated spatial information system for multi-objectiveland-use planning,” Environment and Planning B, Vol. 15(2):205–214, 1988.

5. J.R. Eastman, P.A.K. Kyem, J. Toledano and W. Jin. GIS and Decision making, United Nations Institutefor Training and Research (UNITAR), Geneva, Switzerland, 127, 1993.

6. P. Jankowski. “Integrating geographical information systems and multiple criteria decision-makingmethods,” International Journal of Geographical Information Systems, Vol. 9(3):251–273, 1995.

7. P. Jankowski, G.L. Andrienko and N.V. Andrienko. “Map-centered exploratory approach to multiplecriteria spatial decision making,” International Journal of Geographical Information Science, Vol. 15(2):101–127, 2001.

8. P. Jankowski, T.L. Nyerges, A. Smith, T.J. Moore and E. Horvath. “Spatial group choice: A SDSS toolfor collaborative spatial decision making,” International Journal of Geographical Information Systems,Vol. 11(6):566–602, 1997.

9. R. Kingston, S. Carver, A. Evans and I.Turton. “Web-based public participation geographicalinformation systems: An aid to local environmental decision-making,” Computers, Environment andUrban Systems, Vol. 24:109–125, 2000.

10. J. Malczewski. GIS and Multicriteria Decision Analysis. Wiley: New York, 392, 1999.11. L. Menegolo and R.J. Peckham. “A fully integrated tool for site planning using multi criteria evaluation

techniques within a GIS,” in M. Rumor, R. McMillan, and H.F.L. Ottens (Eds.), GeographicalInformation, IOSA, Amsterdam, The Netherlands, 621–630, 1996.

12. D.A. Mitta. “An application of the analytic hierarchy process: A rank-ordering of computer interfaces,”Human Factors, Vol. 35(1):141–157, 1993.

13. Z.-R. Peng and M.-H. Tsou. Internet GIS: Distributed Geographic Information Services for the Internetand Wireless Networks. Wiley: Hoboken, NJ, 2003.

14. C. Rinner. “Web-based spatial decision support: Status and research directions,” Journal of GeographicInformation and Decision Analysis, Vol. 7(1):14–31, 2003

15. C. Rinner and J. Malczewski. “Web-enabled spatial decision analysis using Ordered Weighted Averaging(OWA),” Journal of Geographical Systems, Vol. 4(4):385–403, 2002.

16. P.S. Roy and S. Tomar. “Biodiversity characterization at landscape level using geospatial-modelingtechnique,” Biological Conservation, Vol. 95(1):95–109, 2000.

17. T.L. Saaty. “A scaling method for priorities in hierarchical structures,” Journal of MathematicalPsychology, Vol. 15:234–281, 1977.

18. T.L. Saaty. The Analytic Hierarchy Process. McGraw-Hill: New York, 1980.19. S. Saran, S. Ghosh, G. Srivastava, P.S. Roy, G. Talukdar and N. Prasad. “Spatial decision support system

for biodiversity conservation prioritization: A web based approach,” Asian Journal of Geoinformatics,259, 2003

20. I.U. Sikder and A. Gangopadhyay. “Design and implementation of a web-based collaborative spatialdecision support system: Organizational and managerial implications,” InformationResources Manage-ment Journal, Vol. 15(4):33–47, 2002.

21. H.A. Simon. The New Science of Management Decision. Harper and Row: New York, 273, 1960.22. R.M. Wallace, Y. Zhang and J.R. Wright. “Distributed system for coastal infrastructure modeling and

assessment,” Journal of Computing in Civil Engineering, Vol. 15(1):67–73, 2001.23. X. Zhu, J. McCosker, A.P. Dale and R.J. Bischof. “Web-based decision support for regional vegetation

management,” Computers, Environment and Urban Systems, Vol. 25(6):605–627, 2001.24. B. Wellar. Science, Applications, Coherence and GIS: Seizing the Moment. GIS/LIS ’90 Proceedings,

Volume 2, California, November 7th–10th.25. M.D. Crossland, W.C. Perkins and B.E. Wynne. “Spatial decision support systems: an overview of

technology and a test efficiency,” Decision Support Systems, Vol. 14(3):219–235, 1995.

Geoinformatica (2007) 11:407–429 427

Page 22: Multicriteria Spatial Decision Analysis in Web GIS Environment

Mr. Harish Chandra Karnatak is postgraduate in Mathematics and Computer applications. Presentlyworking as a Scientist in Geoinformatics Division of National Remote Sensing Agency, Department of SpaceGovernment of India. His area of interest includes Enterprise GIS, Spatial Database Management, RDBMS,GIS analysis, modeling and application development, Multicriteria decision analysis, SDSS and Data mining.He is working on various national level operational/research projects on Web/Enterprise GIS, SpatialDatabase Management and Standardization, Software application development and Geo-spatial Modeling.

Dr. Sameer Saran is postgraduate in physics and is Doctorate in the field of Geoinformatics. He is presentlyScientist at Indian Institute of Remote Sensing. He is working on various projects of Web GIS applicationsand Geospatial Modeling. His other field of interest includes Microwave remote sensing, Process basedModeling, Spatial Decision Support System, Geo databases and Spatial data mining.

428 Geoinformatica (2007) 11:407–429

Page 23: Multicriteria Spatial Decision Analysis in Web GIS Environment

Dr. Karamjit Bhatia is a Reader in Department of Computer Science at Gurukula Kangri University,Haridwar (INDIA). He did M. Phil. in Computer Applications from University of Roorkee, Roorkee (nowIIT, Roorkee). He received his Ph. D. in Computer Science from Gurukula Kangri University, Haridwar. Hisarea of inetest includes distributed systems and image processing.

Dr. P.S. Roy is currently Deputy Director (RS & GIS AA) NRSA, Department of Space Government ofIndia Hyderabad. He received his master degree in Botany and doctorate in ecology. His area of specilizationincludes landscape ecology, biodiversity and environment science, remote sensing and geoinformationapplications. He has published more than 112 research papers in peer reviewed national and internationaljournals. He has supervised 12 Ph. D scholars.

Geoinformatica (2007) 11:407–429 429