Home Articles Integrating GIS and modelbase

Integrating GIS and modelbase

V. K. Panchal, Lokesh K. Sinha, Vinay kanwar and S. D. Mehta

 

Abstract
The geographical applications deals with objects whose not only position in space matter but temporal changes also have to be studied for gaining insight into the futuristic trends. This is achieved through modelling and simulation of the geoprocess. Spatial data management and analysis is handled by Geographic Information system technology efficiently. In spite of this the integration of GIS with modelling system remains a very active research area. In this paper, a loosely coupled system is proposed in this direction.

GIS is the best example of information technology, which captures, manipulates, process and display multisource and voluminous geo-referenced data. With in the last 25 years this technology has encompassed the science and technology of remote sensing, cartography, surveying, geodesy, photogrammetry and of course the 'mother' science to GIS, geography.

Geometric and thematic data about terrain objects stored in a GIS can be kept up to data by using remotely sensed data. Geometric and thematic information can be extracted from RS data by segmentation and classification techniques respectively. As the terrain objects change in time, their representation in GIS should be updated. But, in the situation when simulation of the geo-referenced activities are required for decision making then it is expected the GIS be linked to a modelbase management system. The models for river flood plain landscape combines a purely spatial process, the stochastic pyramid with analytical description on each landform of fluvial landscape or for the rainfall model, which may have submodels like the throughfall, the surface flow, overland flow, stream flow, sub-watershed and sub-surface model

Not only in environmental problems, in scientific and engineering, military domain, assessment also involves forecasting and analysing WHAT_IF scenario, which is an inherently difficult problem in any domain. (Agarwala, 1992; Colombo,1992).

The management of natural resources requires the integration of very large volumes of information from numerous sources. The information technology, and in particular, the integration of database management system, GIS, remote sensing & image processing, simulation & multicriteria optimisation models, expert systems and computer graphics provide effective tools for decision making.

Most environmental and resource management problems like movement potential, natural & manmade hazard zonation, flooding potential, environmental studies, disaster management etc., have an obvious spatial component. Within the domain of environmental modeling this is addressed by spatially distributed models.

The integration of complex & powerful software tools in problem oriented systems provide direct & easy access to gigantic bytes of data. it supports their interactive analysis and helps to display & interpret results in a common graphic user interface.

The multifaceted nature of many terrain problems needs access to a range of models, data and other information. Modeling allows prediction of an expected future state. Modeling and simulation provide a rapid means of investigating the expected response of a system to possible future changes by undertaking the necessary computation which are commonly complex and data intensive.

The necessity to integrate simulation models of these kind of spatial process with GIS technology is well recognised (burrough et. Al. 1988, Densham 1993). These integrated system offer a virtual environment where user can assess the scenario and evaluate various strategies.

The available software for modeling and simulation are very advanced & some of the desired features such as representation of spatial data and provision of expert help, can not be easily delivered using such S/W (young). Just how simulation S/W is to be integrated with GIS software is a subject. Nyerges & goodchild identify S/W integration strategies that range from loosely coupled to full integrated system.

Loosely coupled system:
In a loosely coupled system GIS software is used to construct input files that a simulation program can read. The result of the simulation is then read back into the GIS S/W for display and analysis. The loosely coupled system may be developed using existing technologies, but this integration is lacks in providing.

  • A consistent user interface.
  • A consistent data structure.
  • Support for development & modification of models
  • User interaction during a simulated event.

Tightly coupled System:
GIS user, in this system has access to simulation models through software hooks and or built in macro languages. These integration strategies can provide access to a consistent user interface and data structure, but currently available S/W does not support model development or user interaction during a simulated event. (figure)

Figure 1
Ideally, in a fully integrated system, simulation models, GIS capabilities are part of the same geo-processing S/W. Such a software system should support the construction, execution & manipulation of geographical simulation model in seamless, user friendly environment.

In this paper we have shown how this problem can be tackled using existing technologies till the time a modelbased GIS is available commercially House

 


Enabling Technologies:
To generation a system, that support the creation & execution of geographical models, following technologies are proposed:

A. Modelbase management system:

  1. Modelbase management system (MBMS) stores, manipulate & retrieve models in a manner that is analogous to the management of data within a database management system. A MBMS must incorporate knowledge about which models apply, in which situation and be general enough to manage different model types and multiple views of the same model. If multiple views of a given model can be easily construct, then, a mechanism exists to quickly customise a model to meet the unique requirements of a particular study (Dolk & Konsynski 1984).
  2. S/W tools for modeling & simulation: SIMULA (Dahl et.al 1970) – is the first simulation language & contains some constructs which are embryonic forms of those found in the object oriented programming paradigms; CINEMA: is a language able to handle complex visualisation tasks common in displaying the results of many simulation experiments: MODISM: evolving from SIMSCRIPT has the power of object oriented paradigm and the flexibility of a programming language to simulation .
  3. Tools for calculations: It would be difficult to distinguish tools for scientific calculus from simulation languages, were it not for the fact that the latter provide ready to use instructions to build models, while the former focus on providing a wide spectrum of calculus functions, with simulations being just an application of these functions . MATLAB & MatrixX resemble simulation languages quite closely & can be used to write specific application for the end-user. However, Maple, Mathematics & MathCAD do not allow for this. Any simulation models built using these tools always requires that the user has a working knowledge of a particular simulation software.
  4. Visual Modeling & Simulation environment: A visual modeling environment based on the system theory approach (Wilson, 1984) uses the formalism to represent system as boxes with input, output, states and parameters. Some examples of the tools which use this approach are :

    Simulink: is one of the most successful commercial products which largely owes it success to MATLAB, since itcan be seen as a graphical shell built around the kernel of MATLAB. VisSim (Darnell, 1996) & Extend (Diamond, 1996) are fully graphical tools with the capability to customise the graphical interface to allow interaction with the implemented models.

  5. Model reuse & Integration: Model re-use is important to avoid the duplication of past model development efforts. We should keep in mind that providing integration software & standards is particularly difficult because of
    1. The lack of a common specification for model interface.
    2. The difficulty in extracting the sub-model from an existing mode.
    3. The unavailability of machine independent code.
    4. Frequency with which models are integrated in software packages and can not be used independently.

A number of environmental model bases have been developed to help facilitate mode reuse. Methods for reusing & integrating the existing body of knowledge have been actively investigated over the last five years, mostly using object-oriented techniques. This research is still in stages.

B. Object-Oriented GIS:
Work in object-oriented GIS (Worboys et.al 1990, Raper & Livingstone, 1995) and SDSS (Bennett et.al 1996) illustrate the utility of the object-oriented paradigm in the representation of geo-graphical phenomena. A fully implemented Object-oriented GIS possess a class library that contains standard spatial data structures. These classes may be used through the object-oriented constructs of inheritance & polymorphism (Panchal et.al. 1996). An object-oriented approach is used to provide

  1. Extensibility needed to create new geo graphical moels
  2. Semantic power needed to build complex objects that capture the spatial states, process & relations of geographical systems.
  3. Flexibility needed to develop simulation models that can adapt to the changing states of geographical system.

An examples of an object-oriented GIS is Gothic IGIS from Laser-Scan providing support for active object database. Other object-oriented GIS interface is Geo-media from Intergraph that provides a common user-interface and linkage to various GIS functionality through the use of Visual Basic & C++.

Therefore, the proposed strategy is through the use of object-oriented technology. The interfacing active objects, which are responsible for invoking models developed in MATLAB and carrying the result back to Gothic-GIS/Geo-media can be created in C++. Further, from the analysis of these two GIS software based on Object-Oriented technology, a common interface to GIS functionality and Modelbase can also be developed through the use of visual C++ and the development toolkit of the respective software. In brief, the interface can be shown figure 1.

Alternatively this may be depicted as, taking any object-oriented GIS tool kit:

This strategy, though loosely coupled, in the sense, that it is not an integrated environment, overcome the deficiencies of such a system that it provides a consistent user interface, a consistent data structure, support for development & modification of models, and user interaction during a simulated event.

References:

  • Rizzoli, A. E. and Yound, P. J. (1997) Delivering environmental decision systems. Environmental Modeling & software 12, 2-3.
  • Dolk D. R. and Konsynski, B. R. (1984) Knowledge representation for model mangement systems. IEEE Transactions on Software Engineering 10, 619-628.
  • Colombo, A. G., editor. 1992 Environmental impact Assessment for Developing Countries. Butterworth, Oxford.
  • Kurt Fedra. (1998) Decision support Natural resources Management. International Institute for Applied System Analysis, Luxemburg, Austria.
  • Integration of GIS and Hydrologic Models (1995): Rev. Geophysics. Vol.33 supp.
  • Expert and knowledge-base system(1995): Rev. Geophysics. Vol.33 Suppl.
  • Lucas, L. F. Janssen and Martien Molenaar(1995)Terrain Objects, thie Dynamics and Their Monitoring by Integration of GIS and Remote Sensing. IEEE Transactions on Geoscience and Remote Sensing, 33,3.
  • Wilkinson, G.G. (1996) A review of current issues in the integration of GIS and remote sensing data. Int. J. Geographical Information Systems, 19,1.
  • Bennett, D. a. (1997) A framework for the integration of geo-graphical information systems and modelbase management, Int. J. Geographical information System, 11,4.
  • Dahl, O. J. Myhrhaug, B. and Nygaard, K. (1970) SIMULA 67 common base language. Publication N. S.-22 Norwegian Computing Centre, Oslo.
  • Darnell, K. (1996) VisSim User Manual. Visual Solutions, Westford, M. A.
  • Worboys, M. F., Hearnshaw, H. M. and Maguire, D. J. (1990), Object-oriented data modeling for spatial databases. International Journal of Geographical Information Systems 4.
  • Raper, J., andLivingstone, D. (1995) Development of geomorphological spatial model using object-oriented design. International Journal of Geographical Information Systems 9.
  • Burrough, P. A. Van Deursen, W. and Heuvelink, G. (1998) Linking spatial models and GIS; a marriage of convenience or a blossoming partnership? In procedding of GIS/LIS'88, (Bethesda: American Congress on Surveying and Mapping), pp.598-607.
  • Panchal, V. K., Pathak, B. K., Moorthy, A. S. (1993) Remote Sensing Updates for Data Structure in Terrain Evaluation. National Symposium on Remote Sensing Applications for Resource management, Guwahati, India .
  • Panchal V. K., Pathak, B. K., Moorthy, A. S. (1993) Role of Integrated Surveys in Development of Computer Based Terrain information System. National Symposium on Remote Sensing Applications forResource Management, Guwahati, India.
  • Panchal V. K., Shina, L. K. Prusty, B.G. Chansarkar, R. A. (1996) Object-Oriented Design to Terrain Analysis. Indo-US Symposium-workshop on Remote Sensing & its Applications, Mumbai, India .
  • Panchal V. K., Shina, L. K. Prusty, B.G. Chansarkar, R. A. (1996) Constrains of Satellite Data. Indo-US Symposium-workshop on Remote Sensing & its Application, Mumbai, India .
  • V. K., Panchal Lokesh K. Shina, S. D. Mehta Defence terrain Research, Delh and Vinay Kanwar, CZS College of Engineering, Bhatinda.