Minnesota Department of Transportation
Minnesota Department of Transportation

511 Travel Info

Snowy field


Minnesota Statewide Archaeological Predictive Model

MnModel Home | Data | Procedures | Results | Implementation | Terms of Use | Contact


Quick Links





















Mn/Model Research Design
Appendix E: Introduction to Geographic Information Systems


Mn/Model logo



A Geographic Information System is an organized collection of computer hardware, software, geographic data, and personnel designed to efficiently capture, store, update, manipulate, analyze, and display all forms of geographically referenced information. It incorporates the essential elements of computer cartography and relational databases into one system. The most important characteristic of this system is that every mapped feature is linked to a record in a tabular database and may be related to records in other databases as well. In other words, the GIS fully integrates geographic and tabular data.


This linkage between maps and tabular data makes analysis of geographic data possible. Computer cartography allows representation, but not analysis, of geographic data. Relational database managers do not contain geographic information. Integrating those two technologies in GIS creates "intelligent maps" and the ability to perform spatial analysis. Such analysis may include spatial queries ("where is?"), spatial measurements ("how far?"), or more complex problems (best route, spatial correlation, districting, etc.).


GIS data are organized in layers. Each layer contains only one type of information about the area in question. For instance, one layer might be a soils map with the associated database containing information about a large number of soils variables (texture, pH, permeability, etc.). Another layer may be a map of roads, associated with a database containing road name, road class, number of lanes, pavement materials, date last paved, etc. All of the layers within a given project would represent the same geographic area. These layers can be overlaid with one another to allow the analysis of relationships between the different layers.


This kind of analysis is possible because the GIS is geographically referenced. That is, it is in a real-world coordinate system (Latitude/longitude, UTM, State Plane), allowing accurate overlay of layers containing different data themes for the same geographic area. For instance, a highway project corridor could be overlaid with a soils map, even if they came from sources at two different scales.


GIS Function

Several analytical functions are supported by the integration of geographic and tabular data. The most simple of these is database query. The GIS automates searches for user-specified subsets of the database. GIS is unique in that it allows searches in two different ways. A query may be addressed directly to the tabular data, then displayed on the map ("select all counties with populations over one million and color them yellow"). Alternately, it may be addressed to the map, with the results displayed in tabular form ("show me the population and mean income data for the five counties I have pointed to").


GIS also supports spatial analysis, including automatic computation of lengths and areas, recognition of adjacency, and finding spatial correlations. Spatial analysis can answer questions such as: How many acres of organic soils are within the corridor? What land uses are within 500 meters of the corridor? What kinds of natural features are adjacent to prehistoric hunting camps? On what kinds of soils are you likely to find jack pine growing? What five environmental variables are most often correlated with the presence of an archaeological site?


The most sophisticated kind of analysis that can be performed by GIS is modeling. Modeling has three properties. First, it involves the analysis of complex factors. For instance, modeling fire behavior would involve analyzing forest type and density, amount of dead fuel, slope, current soil moisture, wind speed and direction, and a host of other independent variables.


Second, successful modeling requires that the relationship between factors be understood and clearly defined. The relationship may be expressed with mathematical or Boolean operators. For example, Darcy's Law is a mathematical expression of the relationship between the rate of water flow within an aquifer and the permeability and slope of the aquifer.


Finally, modeling is predictive. By understanding the relationships between a complex of factors and the dependent variable, one can predict the value of the dependent variable in different places or circumstances. Thus, modeling answers such questions as: Where is there a high probability of finding an archaeological site? How long will it take a contaminant from Well A to reach Well B? What forest stands are most likely to burn in this fire season?



GIS is an excellent system for organizing and managing information that is spatially referenced (i.e., can be mapped). It efficiently handles very large databases and maintains links between maps and tabular databases. Beyond its data storage and management capabilities, it provides powerful tools for analysis of spatial data. This includes analytical and modeling functions that are not practical or possible with other methods. It has been successfully applied to the development of archaeological predictive models and is the ideal technology for this project.


Modeling with GIS

Predictive models developed with GIS fall into the general category of cartographic modeling. Cartographic modeling is a general methodology for the analysis and synthesis of geographic data. It applies map algebra to sets of single-factor maps, which are treated as variables that can be manipulated using map algebra functions. With these functions, variables can be transformed or combined into new variables.


The first step in modeling is to define the variables and questions in unambiguous terms. For instance, say the question is how to get to the biggest hill in an area. Defining the question and variables would involve answering the questions: What is a hill? What defines the biggest hill? What factors might determine preferred access to that hill? These questions would be addressed through a series of steps which transform raw elevation data to new layers representing the newly defined variables.


Modeling conventions

The body of data used for cartographic modeling consists of a set of map layers for the given study area. Each layer is a two-dimensional representation (map) of exactly one variable. Two important characteristics of each layer are its orientation (deviation from north) and resolution (size of the smallest mappable unit). Orientation is an important property for modeling, as it may be useful information for modeling other layers, such as aspect or preferred direction of travel. The minimum mapping unit will determine the resolution of the model.


New layers in a model may be created by transformation from other layers. Transformation functions are not unlike algebraic equations, where variables are numerical quantities that are processed by a variety of operations (addition, subtraction, etc.). Map layers are variables that are processed by GIS operations. This may include operations to reclassify zones, to combine layers, to calculate distances and directions, to measure sizes, to characterize shapes, to determine lines-of-sight, to simulate dispersion, etc.

Each map algebraic operation accepts one or more map layers as its input and generates a new layer as its output. Thus, the output from any operation can be used as input to another. Sequences of such operations are called procedures. Such procedures can be used to model complex phenomena such as soil erosion. Nevertheless, their simple components make it possible to express even complex models in a clear and consistent manner. The control script of a model specifies particular operations, identifies the map layers to which they apply, and indicates the order in which they are performed.


Modeling capabilities

The general capabilities of cartographic modeling facilitate the interpretation of geographic data. Interpretation is a process in which facts (data) are translated into more useful facts (information). Geographic data interpretation (as opposed to simple representation) is the major distinguishing characteristic of GIS.


The data interpretation process involves extracting relationships or meanings that are implicit in a set of data and expressing these in explicit form. An example would be deriving slope and aspect from a map of elevation. This transformation of data is facilitated by the fact that map features are represented not by lines or symbols, but by numerical values, and that these values are directly associated with specific locations. The use of numbers by the GIS makes it possible to transform them with mathematical functions. The data-transforming functions of a GIS model are classified as local, zonal, incremental, and focal operations.


Local functions are those that compute a new value for every location as a function of one or more existing values associated with that location. This is usually done by applying a mathematical function to each location's value on one or more existing map layers. An example would be computing population growth by subtracting 1980 population from the 1990 population for the same location.

Zonal operations compute a new value for each location as a function of the existing values from all locations in a second layer that have the same value in the first layer. For example, it could determine the most extensive soil type found within the zone of a forest cover type from another layer.


Incremental operations characterize each location as an increment of one-, two-, or three-dimensional cartographic form. The size and shape of these increments are inferred from the value of each location relative to those of its adjacent neighbors. An example would be the computation of aspect (the compass direction of steepest descent) from a map of elevation.


Focal operations are those that compute each location's new value as a function of the existing values, distances, and/or directions of neighboring (but not necessarily adjacent) locations. The distance relationships may be defined by such variables as physical separation, travel costs, or inter-visibility. Examples might include computation of the most efficient direction of travel to a specific point or the total value of land within a location's defined neighborhood.


Modeling Techniques

Among the wide array of modeling techniques, a distinction can be drawn between descriptive modeling and prescriptive modeling. Descriptive modeling is concerned with 'what is' or 'what could be.' Prescriptive modeling is concerned with 'what should be.'


Descriptive modeling techniques may either analyze or synthesize geographic data. Analysis decomposes data into finer levels of meaning, while synthesis recomposes data for use in particular contexts. Analysis generally involves the characterization of either position or form. For example, position might be expressed as travel time, while form is expressed as slope. Analytical modeling techniques tend to be associated with applications that are oriented towards the acquisition of objective knowledge.


Synthetic modeling techniques tend to be associated with applications involving the exercise of subjective judgment. Usually these techniques call for the use of an operation to combine map layers that represent major factors affecting a question or decision in a way that specifies the relative importance of those factors. An example might be an operation to indicate the relative importance of each of several zones on a layer of observable site conditions.


In either case, the formulation of a descriptive cartographic model can generally be accomplished best not by proceeding inductively from existing data to envisioned results but by proceeding deductively from envisioned results to the data from which they will ultimately be derived. In other words, define the problem or question, determine what you need to know to answer the question, then determine the best source and format of those data.


Prescriptive modeling techniques are found in more active or deliberate forms of decision making and problem solving. These techniques are generally associated with some form of geographic allocation, i.e. selecting locations in order to accomplish specific objectives. This process generally involves the statement of a problem, the generation of solutions to that problem, and the evaluation of those solutions. The statement of the problem is essentially a descriptive task, consisting of defining criteria for its solution. For instance, if the problem is to locate a landfill, one of the criteria for its solution might be a stated level of soil impermeability.


To generate a solution to the problem, the statement of criteria must be transformed into a set of locations that satisfy the criteria. The solution will be a selection of sets of locations that exhibit all of the identified criteria. Finally those solutions can be evaluated in terms of both the pre-defined criteria and of other issues, such as political issues, beyond the original problem statement.



The origins of cartographic modeling can be traced to those of manual map overlay. Digital overlay and application of map algebra to large databases make cartographic modeling with GIS both more powerful and more accurate than manual overlay. As opposed to relational database or object-oriented models, cartographic modeling is especially good at dealing with distances and other spatial concepts such as narrowness, enclosure, spottiness, interspersion, striation, and so on. Cartographic models can be developed for high-level applications in a variety of fields. The most successful models will be generated by those who ultimately put them to use (i.e., archaeologists, geologists) rather than by those who are more technically oriented (i.e., programmers, software developers). Moreover, unlike a few years ago, commercially available software provides both a full set of map algebraic functions and integration with powerful statistical software.


References Cited

Dobson, J. E.
1991 Exploring geographic analysis. In 1991-92 International GIS Sourcebook, edited by H. D. Parker, J. Stutheit, J. E. Dobson, J. R. Hughes, and G. K. Rodcay, pp. 335-337. GIS World, Inc., Fort Collins, CO.


Environmental Systems Research Institute
1992 Cell-based Modeling with GRID: Analysis, display, and management, Second edition. Environmental Systems, Redlands, CA.

Landis, J. D.

1993 GIS capabilities, uses, and organizational issues. In Profiting from a Geographic Information System, edited by Gilber H. Castle, III, pp. 23-53. GIS World, Inc., Fort Collins, CO.


Maguire, D. J.
1991 An overview and definition of GIS. In Geographical Information Systems: Principles and Applications, edited by D. J. Maguire, M.


F. Goodchild, and D. W. Rhind, pp. 9-20. Longman Scientific & Technical, London.

Openshaw, S., and C. Brunsdon
1991 An introduction to spatial analysis in GIS. In 1991-92 International GIS Sourcebook, edited by H. D. Parker, J. Stutheit, J. E.


Dobson, J. R. Hughes, and G. K. Rodcay, pp. 401-404. GIS World, Inc., Fort Collins, CO.

Star, J., and J. Estes
1990 Geographic Information Systems: An Introduction . Prentice Hall, Englewood Cliffs, NJ.


Thompson, D.
1991 Geographic Information Systems: an introduction. In 1991-92 International GIS Sourcebook, edited by H. D. Parker, J. Stutheit, J. E. Dobson, J. R. Hughes, and G. K. Rodcay, pp. 338-341. GIS World, Inc., Fort Collins, CO.


Tomlin, C.D.
1991 GIS in Cartographic Modeling. In Geographical Information Systems: Principles and Applications, edited by D. J. Maguire, M. F. Goodchild, and D. W. Rhind, pp. 361-374. Longman Scientific & Technical, London.


Wellar, B.
1993 GIS fundamentals. In Profiting from a Geographic Information System, edited by Gilber H. Castle, III, pp. 3-21. GIS World, Inc., Fort Collins, CO.





MnModel Orange Bar Logo


MnModel was financed with Transportation Enhancement and State Planning and Research funds from the Federal Highway Administration and a Minnesota Department of Transportation match.


Copyright Notice

The MnModel process and the predictive models it produced are copyrighted by the Minnesota Department of Transportation (MnDOT), 2000. They may not be used without MnDOT's consent.