Minnesota Department of Transportation
Minnesota Department of Transportation

511 Travel Info

Jeffers petroglyphs


Minnesota Statewide Archaeological Predictive Model

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


Quick Links





















Draft Research Design for the Development of a High Probability Predictive Model for Identifying Archaeological Sites

This is the original Mn/Model Research Design, submitted in 1996. Research methods evolved over the course of the project. These changes, and the procedures used to achieve the final results, are documented in the Final Report.


Mn/Model Logo

MnDOT Agreement No. 73217
BRW Project No. 2999/3501


Prepared for:
Minnesota Department of Transportation


Funded by:
Federal Highway Administration
Intermodal Surface Transportation Efficiency Act (ISTEA)


Prepared by:
BRW, Inc.
700 South Third Street
Minneapolis, Minnesota 55415


Foth & Van Dyke
Leech Lake Heritage Sites Program
Mississippi Valley Archaeology Center
University of Minnesota - Twin Cities

July 10, 1996



Table of Contents




A. Basic Data Accumulation


B. Developing the Initial GIS Model



A. Supplementing the GIS Database

B. Developing the Third and Fourth Dimensions of the Model



A. Model Demonstration and Fine Tuning

B. Final Implementation and Model Operating Characteristics




APPENDIX A: What is an Archaeological Predictive Model?

APPENDIX B: Modeling Theory and Assumptions

APPENDIX C: Archaeological Field Survey Methodology

APPENDIX D: Geomorphology/Geology

APPENDIX E: Introduction to Geographic Information Systems

APPENDIX F: Using a Cultural Resources Predictive Model

APPENDIX G: Glossary of Terms





The Minnesota Department of Transportation (MnDOT) spends over one million dollars each year looking for and evaluating archaeological resources threatened by proposed transportation routes and other construction and maintenance activities. Presently, there is no formal planning mechanism that aids in managing these resources by indicating where unrecorded sites are likely to be located and where survey efforts should be focused. Rather, a search for archaeological resources begins only after or near the completion of preliminary construction designs. When archaeological resources are discovered, unexpected costs occur due to work delays and the expense of examining the resources. In response to this situation, cultural resource managers at MnDOT began exploring more efficient and proactive approaches to fulfill the requirements of Section 106 of the National Historic Preservation Act of 1966 (as amended). Section 106 mandates that any project using federal funds or permits undergo a review process that includes identifying and evaluating archaeological sites in the project area. As a result of this efficiency review, MnDOT cultural resource managers sent out a request for proposals to develop a statewide predictive model for archaeological resource locations. The project is sponsored by MnDOT using funds from the Federal Highway Administration's Intermodal Surface Transportation Efficiency Act (ISTEA).


This research design presents a plan for achieving this goal. This section provides an operational outline of the team's approach to the project. Extended discussions of more theoretical and technical aspects of the project are reserved for appendices at the back of this document - what is an archaeological predictive model (Appendix A), the theoretical basis of the predictive model selected (Appendix B), a standard methodology for archaeological field survey (Appendix C), developing geomorphological/geological scopes of work (Appendix D), and Geographic Information Systems (GIS) (Appendix E). Appendix F demonstrates how the team's user-friendly predictive model will work from an operator's perspective. Appendix G is a glossary for the technical terms used in the research design.

The team's research design for archaeological predictive modeling utilizes an empirical approach. To develop the model, we will examine patterns exhibited by rigorous archaeological surveys. Our model will use state-of-the-art computer technology and proven methodologies. Our strategy is simple: using Geographic Information Systems (GIS), we identify locational patterns observable in archaeological samples from throughout Minnesota. With this method, archaeological site locations are the dependent variables. They are predicted by combinations of environmental variables, which are the independent variables. This has been successfully achieved in several studies in the United States and Europe by members of the team. Geographic Information Systems allow rapid and detailed mapping and analysis across broad regions. The final product in both computerized and paper form will be an easy-to-interpret and useful product that indicates areas of low, medium, and high potential for archaeological sites.


The field and laboratory procedures proposed here are based upon the requirements of this kind of modeling approach, which has been developed over the last 15 years and is described in Appendices A and B. These requirements are: 1) data acquired through probabilistic means, 2) a carefully designed model-testing program, 3) a user-friendly operator interface, and 4) a model that can be updated, allowing it to grow in accuracy throughout the lifetime of its use. Given these requirements, the three-year project proceeds by necessity through three phases: 1) basic data accumulation and model development, 2) model enhancement, and 3) model refinement and final implementation.


Since some segments of each phase will proceed more rapidly than others, each phase will not be equal to one year. Rather, the first phase should take longer than the second, which will be more brief than the third. This distribution of effort is reflected in the length of discussion of each phase below.


Return to the top



During the first phase of the project: 1) we will gather basic data for both independent (environmental characteristics) and dependent (archaeological events) variables by converting existing data, conducting new field surveys, and examining Minnesota Statewide Archaeological Survey manuscripts, and 2) we will develop an initial model. Each of these tasks is briefly described below.

A. Basic Data Accumulation

The most accurate cultural resource predictive models are based on field information gathered with probability-based search procedures. These models are more accurate because they use randomly located field units to reduce bias. Unlike inductive models, they do not assume that the distribution of known archaeological resources is necessarily a reliable guide to the distribution of all resources of this kind. The majority of known sites, for example, are large (and thus prominent) base camps located near large lakes and rivers. Investigations of these sites where good season-of-occupation information is available demonstrate that they were occupied only during the warm weather months of the year. Where are the late fall, winter, and early spring sites?


The issue is complicated further by the great mobility of early Archaic and Paleoindian people in the state before 5,000 years ago. These people camped briefly in many diverse environments in Minnesota as they traveled over large areas in search of large game animals and other food resources. Their sites are nearly always small. As a result, they remain underreported, for they tend not to be noticed or considered important enough to report if found by landowners. Finally, by concentrating only on the distribution of known sites, we fail to learn the characteristics of places that were not occupied, a source of information of equal importance to resource planning as the nature of places that were occupied. To ensure a highly accurate predictive model for all parts of the state, it is essential, then, that the model be based on information gathered using probability-based search procedures. Furthermore, the model must apply equally well to parcels of land that most likely do not have cultural resources as to those that do.

The State of Minnesota is fortunate in this project to already have a very large if still untapped data base, the late 1970s Minnesota Statewide Archaeological Survey funded by the Legislative Commission on Minnesota Resources (LCMR). That project investigated more than twenty counties using probability-based search procedures. Our team's approach will build on that data base and on the results of a number of more recent large-scale, rigorous Cultural Resources Management (CRM) surveys, such as pipeline and Minnesota Historical Society Trunk Highway archaeological reconnaissance surveys. In spite of the theoretical soundness of a probabilistic approach, it is often not practical for financial reasons. It is much more expensive and time consuming to gather new probability-based data than it is to use the location of known sites as a guide to the location of unknown sites. The original statewide survey already provides much of our needed data. However, that data base is not without its problems. A variety of strategies of data accumulation were adopted by the Minnesota Statewide Archaeological Survey, and some sampling errors occurred. As a result, a small but more rigorous survey will be conducted in this phase of the project. Besides providing additional field data, it will allow an assessment and standardization of the original data base so that it can be put to the predictive use for which it was intended. Designing a survey of this nature and scope involves decisions concerning sampling strategy, and field survey and geomorphological testing procedures.


Return to the top


Predictive models of archaeological resource location can be based upon a variety of basic landscape variables that differ widely in scale and nature. Some of these variables are continuous, such as slope and aspect, others are discrete, such as landform or vegetation types. A primary goal of a modeling project like this one is to determine which combination of landscape variables produces the highest accuracy in predicting resource presence or absence. Effective combinations may vary both locally, within different parts of the landscape, and regionally, between different environmental and geographic regions. Thus, the model will be composed of a number of sub-models, each specific to an identifiable landscape unit (such as river valleys) within a given environmental region. This project will begin by sampling discrete landscape units within several regional environments. The landscape units selected will maximize comparison with data derived from the Minnesota Statewide Archaeological Survey.

The environmental regions chosen for this project are the nine archaeological regions identified for the State of Minnesota by the National Register Archaeologist. The regions are defined based on the surface hydrology of the state, that is, on the nature of its lake and stream systems. Because of their importance as sources of food and as transportation routes in the prehistoric and early historic periods, these systems are thought to have structured subsistence and settlement options. If this conclusion is correct, archaeological resource locations should be more homogeneous within regions than between regions, and each region could serve as the geographic framework for a sub-model. The regions are (from southwest to northeast): the Southwest Riverine, Prairie Lake, Southeast Riverine, Central Lakes Deciduous, Central Lakes Coniferous, Red River Valley, Northern Bog, Border Lakes, and Lake Superior Shore.

In the Phase 1 sampling design for the summer of 1995, which is outlined below, landscape types within three of these regions will be the focus of field survey. This procedure will ensure that landscape types used for stratification in the Minnesota Statewide Archaeological Survey are equally well represented within each region. The relative importance of landscape variables can then be contrasted between environmentally different regions. By the end of Phase 1, a tentative conclusion will be reached concerning the landscape variables, or combination of variables, that seem most useful for model development. The integrity of this framework will then be tested and refined in Phases 2 and 3.

The Phase 1 field survey portion of the project will be conducted during the warm weather months of 1995 and 1996. During the first summer, the team will survey characteristic landscape types in Nicollet (Prairie Lake region), Stearns (Central Deciduous Lakes region), Becker (Central Deciduous Lakes and Central Coniferous Lakes regions), and Beltrami (Central Coniferous Lakes region) counties. These samples will be taken in river valleys, around lakes, and in other landscape units that occur in each county. Portions of United States Geologic Survey (USGS) map quarter-quarter sections representing these landscape types will be selected using a stratified random sampling procedure and then field surveyed for the presence or absence of archaeological resources. Three field crews will conduct the survey. A Mississippi Valley Archaeology Center (MVAC) crew will survey the Nicollet County parcels; a BRW crew will survey portions of Stearns County and the eastern region of Becker county; and a Leech Lake Heritage Sites Program (Leech Lake) crew will survey parcels in Beltrami County. The goals of this initial field survey are: 1) to gather data sets that can be contrasted to reveal the relative predictive power of landscape variables within different environmental regions, and 2) to refine the definition of landform variables so that the variables used in subsequent phases of the project make the most archaeological sense. Since the sampling frames in different regions will vary in size of USGS section surveyed, in portion of county surveyed, and in number of sections surveyed because of differences in landscape variability across counties and ease of survey (e.g., forested areas are more time consuming to survey than are cultivated fields), documented sites and environmental data will be mapped in the field at appropriate resolutions for each region.

The first summer's sampling strategy is designed 1) to ensure comparability of survey results with those of the earlier Minnesota Statewide Archaeological Survey, and 2) to ensure that all landscape types selected within the sampling frame (see below) will be tested regardless of their suspected site potential. The latter goal is as important in this project as the first, for it is as important to state with equal confidence in a cultural resources predictive model where sites will most likely not be found as where they most likely will be found. The strategy adopted is designed to sample an equal number of different landscape types in a survey region. Since the landscape types present in the three environmental regions that will be surveyed the first summer vary slightly, a subdivision process will be employed based upon basic adjacent-to-water and away-from-water division used in the Minnesota Statewide Archaeological Survey. Each of these basic divisions will be further subdivided in accordance with the terminology used in the Minnesota Statewide Archaeological Survey and the landforms present in each survey region. For example, in Stearns County adjacent-to-water will be divided into Stream/River, Lake, and Wetland, Lake into "Larger than 40 acres" and "Smaller than 40 acres" (a Minnesota Statewide Archaeological Survey division), and so on. Besides providing a more detailed sample of regional landscape types than in the Minnesota Statewide Archaeological Survey, the strategy ensures comparability with those earlier surveys, for survey results can be compared at different levels of landscape scale. (Minnesota Statewide Archaeological Survey surveys were conducted using varying landscape scales, the most basic of which were simply "adjacent to" and "away from" water.) A detailed rationale and description of this sampling strategy will be provided in the first summer's field report (Mn/Model Interim Report # 1), which will be prepared in the fall of 1995.

During the second summer, team archaeologists will conduct field survey in Wabasha (Southeast Riverine region), Wright (Central Lakes Deciduous region), and Cass (Central Lakes Coniferous region) counties. A Mississippi Valley Archaeological Center crew will survey Wabasha County, a BRW crew Wright County, and a Leech Lake Heritage Sites Program crew Cass County. The 40 acre survey parcels the second summer will be selected randomly, with no stratification, by GIS using a random points procedure. The goals of this survey are 1) to compare the results of a completely random survey to those of the geographically stratified, and therefore potentially biased, survey adopted the first summer (to standardize Minnesota Statewide Archaeological Survey data), 2) to provide data to test preliminary predictive models based on Minnesota Statewide Archaeological Survey data, CRM survey data meeting defined criteria, and 1995 Mn/Model field survey data, and 3) to estimate the a priori probability of finding an archaeological site. A priori or "by chance" probabilities are necessary to rate the performance of any predictive model. Wabasha, Wright, and Cass counties have been chosen for survey because Minnesota archaeologists have more data from their environmental regions than other regions in the state. Comparing our model data to the existing information about archaeological sites in those regions will ensure a more severe test of the preliminary models than would surveys in little known regions of the state (where discrepancies could be explained away as the result of atypical site location patterns). A detailed rationale and description of this sampling strategy will be provided in the second summer's field report (Mn/Model Interim Report # 2), which will be completed in the fall of 1996.


Return to the top

Field Survey

In a project of this magnitude and intent, archaeological and geologic field procedures must be systematic and uniform so that reliable data are collected. The Mississippi Valley Archaeology Center at the University of Wisconsin-La Crosse, one of the premier field teams in the Midwest, will be responsible for the field survey portion of this project. Archaeologists at MVAC will 1) coordinate statewide field procedures and activities, and 2) provide appropriate quality control for field procedures.

Each field crew institution (BRW, Leech Lake, and MVAC) will coordinate with the project Research Director in selecting appropriate landscape units in their areas of survey during the first summer's survey. This process will involve a review of known sites and surveys in each area, an assessment of the accessibility of potential survey areas, and the identification of archaeologically significant landscape units in each area. Following the selection of appropriate survey plots, MVAC will establish detailed survey standards and protocols in consultation with the MnDOT archaeologist, National Register Archaeologist, the Office of the State Archaeologist, and survey crew coordinators based on all pertinent topographic, archaeological, and logistical aspects of the selected survey frames (i.e. landscape units). Shortly thereafter, a pre-field orientation and procedures workshop will be conducted involving all survey supervisors, crew chiefs, and coordinators in order to ensure appropriate strategies, techniques, and standards. Each crew will then obtain landowner permission for their areas of survey, and conduct initial field identification and reconnaissance of sample plots.

Field surveys will be conducted by a five-person crew and a crew supervisor following procedures that will be outlined in a field manual prepared by MVAC. Surface reconnaissance will be conducted by visually inspecting disturbed ground, such as cultivated fields, by walking in straight rows 15 meters apart. Survey plots containing areas covered with vegetation will be surveyed by hand-excavating shovel tests at 15 meter intervals. The contents of these holes will be pressed through 1/4 inch mesh hardware cloth to locate artifacts. Depth of shovel testing will vary according to landscape type, as described in the field manual.

All artifacts encountered along a transect, with the exception of large pieces of fire-cracked rock, will be saved for examination in the laboratory. Supervisors will then be responsible for the determination and mapping of site boundaries within areas that result in the discovery of artifacts, and for completing the appropriate state site forms for newly discovered sites. All data recording will be done on standardized forms using field computers where appropriate. Laboratory methods will include cleaning and identifying all artifact samples collected, plotting site locations on USGS maps, and analyzing the data. All data forms and other appropriate information will be sent to MVAC for checking and additional processing. The data forms and site maps will then be sent to BRW for final analysis and entry into the predictive modeling computer program.

Return to the top

Geomorphic Methods

The two major contributions geologists and geomorphologists can make to this project are to interpret: 1) the age of strata and surfaces, and 2) the likelihood of preservation of archaeological resources within strata. The first contribution helps archaeologists determine, for example, which land surfaces could have been occupied by people in the past (the first people in Minnesota probably entered the state in the second half of the twelve millennium B.C., so older buried surfaces can be ignored in surveys). The second contribution helps archaeologists determine the probable presence and significance of buried sites. (Sites in retransported sediments are disturbed and, therefore, less significant according to the National Register of Historic Places [National Register] criteria than those in undisturbed sediments.) Having these third (depth) and fourth (time) dimensions as GIS thematic layers is an essential and distinctive feature of the team's approach to predictive modeling.

Foth & Van Dyke geomorphologists will prepare a detailed high resolution layer of landform-sediment assemblages to be inserted into a GIS format. Attributes of this layer will include landforms within the major river valleys mapped and, for each of these, general sediment sequences, age estimates, and surface and buried potential for archaeological sites. The Minnesota Department of Natural Resources (DNR) statewide GIS surface mapping projects will be incorporated when appropriate and possible, providing more coarse resolution landform information for uplands and river valleys not mapped by Foth and Van Dyke. Field work is necessary to document the underlying geology and to correlate surficial patterns recognized on soils maps, vegetation maps, or aerial photographs within the context of the GIS database. Proper documentation through field work is an essential control in developing an accurate predictive model.

Field work will include describing undisturbed core samples collected when feasible from uplands to bottomlands along predetermined transects within each of the eight MnDOT districts. MnDOT drill-rig teams will be utilized when cores are required within existing highway right-of-ways; otherwise, a combination of a Giddings soil-probe and contracted drillers will be used on private property. Selected organic materials will be collected and radiocarbon-dated to help determine the absolute ages of buried horizons and bracket the ages of land surfaces. Geomorphology will also help bracket the ages of strata and land surfaces based on cross-cutting relations. (For example, a terrace inset beneath a higher terrace is younger than the higher terrace because of the principle of down-cutting relations in a stream system.) Approximately 100 radiocarbon dates will be included in the geomorphic study to build a time framework for the predictive model.

Return to the top

Incorporating Minnesota Statewide Survey Data (1977-1980)

Although the statewide survey was in part both procedurally and statistically flawed, it still has the potential to provide a large data base that can supplement new survey data. Potentially, there are three main problems in interpreting statewide survey data. First, a 50 meter (m) transect and shovel-testing interval was used, which may mean that fewer sites were encountered than would happen using today's standard 15 m interval. Since many Minnesota archaeological sites are very small, this seems a likely possibility. Second, possible statistical errors occurred in selecting quarter-quarter sections for survey (although the initial selection of quarter-quarter sections was random, the procedure used to stratify some counties may have been biased). And third, the absence of a standard stratification procedure means that most survey data are not directly comparable and different proportions of kinds of landforms were surveyed. This data base will be reexamined by the project statistician, and analyzed using variables and criteria established for the first summer's survey mentioned above to determine the probability that smaller sites are underrepresented in the results of the Statewide Archaeological Survey because transects were placed 50 m apart. Locations of sites and randomly located points in quarter-quarter sections surveyed during the statewide survey will be determined and incorporated into the GIS database.

Since the State Historic Preservation Office (SHPO) is presently entering all known site locations (and some other information) in the state into a GIS-format recording system, the Mn/Model project team will not have to record this basic information for Statewide Survey sites. The Mn/Model team will spend its effort in reviewing reports and collections to establish the context of these sites, that is, the types of archaeological components and materials present at these sites. These data will be added to the SHPO's GIS files. The SHPO project team has agreed to give priority in their recording to those counties that were part of the Minnesota Statewide Archaeological Survey in exchange for enriched site information. This joint effort will benefit both projects.

Return to the top

B. Developing the Initial GIS Model

The end-product of this phase of the project will be the development of an initial but usable predictive model based on GIS and the assumptions and principles discussed in the above paragraphs and in Appendix A. This predictive model, based on the large data base provided by the new surveys, selected CRM data, and the statewide survey, will provide a solid GIS-based planning tool. Among the details involved in establishing this model are: 1) the selection and conversion of digital environmental databases, 2) the definition of measurement standards for the independent and dependent variables, 3) the addition of the locations of known prehistoric and early historic sites and random points in negative survey areas, 4) the statistical analysis of the relationships between sites, random points and environmental variables, and 5) the development of GIS procedures and applications to manage the model.

At least 12 environmental databases (more for most counties) will be incorporated in the GIS model in this phase. From these 12 layers more than 40 variables will be derived. Statistical analyses will reduce the number of variables to less than 10 in the final models. The predictive model is intended to forecast the spatial location of both prehistoric and contact period (A.D. 1650-1821) sites, for they share many features, including common locational factors and the rarity of visible, standing structures. In addition, contact period historic sites are often either not mentioned or are only vaguely mentioned in archival records and, as a result, are discovered or must be located through field search procedures. No attempt is made in this project to model the location of post-1837 historic sites, for the types of post-contact period sites present and the reasons for their location are very different from those of earlier sites.

Once the project commences, the project team will immediately begin to carry out these tasks in a pilot project that will involve one county in the Statewide Survey. The pilot project will allow us to work out data conversion methods, definitions of variables, and modeling procedures on a small database. Once our project methodology has been refined through several iterations with the pilot county, it can be applied to entire regions. By the end of the phase (in the fall of 1996 at currently projected funding rates), additional information from the two seasons of field survey and all counties included in the Statewide Survey will be incorporated into the model.

This preliminary model will: 1) be based on a large sample of probability-based field information, 2) operate at a high level of resolution (30 meters) within the local landscape, 3) be fine-tuned to environmental differences between major archaeological regions, 4) apply to both early historic (contact period) and prehistoric period archaeological resources, and 5) indicate low, medium, and high potential for the presence or absence of archaeological resources. The pilot model will be complete enough to contribute to decision making at many levels of cultural resource management.

Return to the top


The goals of the second phase of the project are: 1) to add additional archaeological and environmental data to the GIS data base, 2) to add information about major river valley sediments and the paleoclimate of Minnesota to the GIS data base, and 3) to explore procedures for detecting the presence or absence of sites in deeply buried alluvial sediments. These tasks enhance the model because they add vital information and provide multiple opportunities for assurance control.

A. Supplementing the GIS Database

Data from CRM surveys conducted in Phase II counties will be incorporated into the GIS database in this phase. These new data will further expand the data base for each region and increase the accuracy of the model. Environmental information from other archaeological sites not collected within probabilistic or 100 % frameworks will also be closely examined to better understand the relationships of these variables to the presence or absence of archaeological resources in the various environmental regions of the state. Blocks of data from all of these sources will be set aside to test the predictive power of the models developed for different regions. This process will allow the team to begin to narrow the number of variables incorporated into the model, which in turn will make the model easy to use.

During Phase II, additional environmental data will be added from sources that will not be ready during Phase I. These include the river valley landform-sediment assemblages that will be mapped by Foth and Van Dyke, high resolution soils data for many counties, and other statewide environmental layers being completed by state agencies. These data sources will allow the development of new, in most cases high resolution, landscape variables to improve model performance.

Return to the top

B. Developing the Third and Fourth Dimensions of the Model

A major flaw of nearly all large-scale predictive models of archaeological site location is the failure to model the location of buried sites, a type of archaeological site often encountered in river valley projects, such as bridge repairs or construction, and one that is expensive to excavate. The BRW project team is committed to developing a three-dimensional predictive model of archaeological site location, and believes that models that lack this third dimension are severely flawed for planning purposes. Of equal importance in cultural resource management is understanding the fourth dimension of the age of strata and surfaces in alluvial deposits. Not all buried sites are of equal National Register significance. Very old Paleoindian camp sites, for example, tend to be small with light artifact scatters. However, they are highly significant because of their age and rarity in the state. Modeling the fourth, or temporal, dimension of buried deposits is a means of providing information to the SHPO that will help them assess site significance. As mentioned earlier, approximately 100 radiocarbon dates will be included in the geomorphic study to build an alluvial valley temporal framework for the predictive model. The geomorphic information gathered in Phase 1 will be used in Phase 2 to add depth and time dimensions to the model. These dimensions will be in the form of GIS thematic layers that map the distribution of sediments in major river basins that may or may not contain buried archaeological sites that retain contextual integrity.

The cores collected during geomorphic drilling in Phase 1 will also be used to help develop cost-effective procedures for detecting archaeological material in cores. An example of an artifact detection procedure is the identification of stone microchips (chert dust) in cores.

As a companion to developing a temporal dimension within river valleys, University of Minnesota paleoclimatologists will provide information to the GIS team in Phase 2 that will allow the modeling of the climate (temperature/precipitation) and vegetation of Minnesota for the past 12,000 years. Data will be contained in a series of files for selected weather stations in the state. The team will use a paleoclimate reconstruction tool recently developed by Dr. Reid Bryson at the University of Wisconsin-Madison. Their paleoclimatic models will be linked to eight different types of pollen, from pollen profiles, to model climate/vegetation shifts across the state at a resolution of every 1000 years. Since it is assumed that prehistoric people in Minnesota settled near critical resources, such as water and shelter, it is necessary to know in predictive modeling how these resources shifted through time.

Return to the top


Major tasks of the third phase of the project are: 1) a demonstration of the model's use in actual CRM projects, 2) fine-tuning the model to higher resolution environmental data where available, 3) training sessions for persons who might operate or use the model's output, and 4) presentation of the user-friendly GIS-based model to the state agency that will manage it. These tasks are briefly described below.

A. Model Demonstration and Fine-Tuning

The value of the predictive model will be demonstrated in Phase 3 by applying the software to a project in each of the eight MnDOT districts in consultation with district engineers. These demonstrations will not only serve an important educational function within the context of the actual planning process, they will provide additional data with which to refine the model. Because of the variety of environmental zones in Minnesota, the final model will be a set of models, each of which will be applicable to a different part of the state. The choice of MnDOT planning districts as a means of distributing demonstration projects will ensure that a diversity of environments are sampled.

This aspect of the project will also concentrate on identifying the characteristics of buried sediments that are predicted to contain or not to contain archaeological sites using the geomorphological methods described in Phase 1. A review of the relevant literature will explore 1) the most efficient and cost-effective means of detecting sediments that are likely or not likely to contain buried sites (such as the use of simpler, less expensive coring procedures in less complex alluvial valley situations), and 2) characteristics of Holocene sediments in each of the SHPO's environmental regions that affect the presence or absence of buried sites (such as the destruction of Early Paleoindian sites along the Lake Superior shoreline [Anfinson Region 9] by the Marquette glacial readvance). Testing for the presence or absence of buried sites, as predicted by the GIS Phase 2 model, will occur in selected demonstration projects.

Return to the top

B. Final Implementation and Model Operating Characteristics

A tuned, predictive model will be presented to MnDOT at the end of this process that is: 1) GIS-based, 2) capable of being easily incorporated into the planning process, 3) user friendly and based on simple Windows software displays, and 4) updatable by MnDOT to allow its performance to improve over time. Training seminars will be held at this time with potential users and operators of the model to ensure its optimum effectiveness.

As explained in Appendix F (Using the Predictive Model), the use of the model will be fairly simple. The model will be released for use with ArcView 3 software. This will include a customized Graphic User Interface (ArcView application) developed for this project. The package delivered to the end user will include the ArcView application, environmental layers used to develop the model, a layer of the SHPO archaeological sites, and a layer or layers of the model results. These model results will be in the form of maps, color coded to indicate areas of high, medium, and low probability of finding an archaeological site. The data and ArcView application can be used on any PC, UNIX, or Macintosh computer running ArcView 3.

In addition to the electronic version of the predictive model, several paper versions will be produced to facilitate its use in a variety of situations. Although the exact form of these versions will depend on the structure of the final computer-based model, several possibilities are mentioned here to illustrate their general nature. A common product of archaeological site predictive modeling projects is a set of sensitivity maps that display the distribution of areas of high, medium, and low site potential using a simple color code. These maps could be produced, for example, at the county level. These and other possible formats will increase the utility of the model for all land management agencies in the state.

Return to the top


It is the goal of this team to produce statewide, computer-generated, archaeological predictive models that indicate medium and high sensitivity zones representing a high level of accuracy compared to chance in indicating site presence, when tested against known samples of archaeological sites. Our familiarity with the archaeology of Minnesota gives us confidence that archaeological sites will exhibit strong locational pattern from region-to-region. Nearly a century of archaeology has shown this.

The team also firmly believes in the validity of the proposed methods and general approach to model development, because they represent the combination of state-of-the-art techniques and technologies that have proven to be successful in other regions of the world. Consequently, the team's model will possess levels of precision such that a small percentage of the state will be mapped to the zones of medium and high archaeological sensitivity.





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.