Lecture 13 : GIS Data Modeling (link to Powerpoint file)

Lab 14: Predictive Modeling

Homework 7: Modeling Critique

GIS in Archaeology

GIS Modeling

Introduction

What is a model?

A model is a simplified representation of reality

GIS Models Variations

Descriptive vs. Prescriptive models

Descriptive models describes how the data are

Prescriptive models predict how the data will be under certain circumstances

Deterministic vs. Stochastic models

Deterministic models has completely defined parameters

Stochastic models use and element of randomness in the modeling process

Static vs. Dynamic models

Static models show the state of the data at one moment in time

Dynamic models deals with how the data change over time

Inductive vs. Deductive models

Inductive models use known data to arrive at predictions

Deductive models start from theoretical ideas to arrive at predictions

Raster vs. Vector Models

Raster or Vector, how to decide

Generally the type of model used is largely a function of the types of data sources for the underlying data to be modeled

Vector Models

Good for modeling phenomenon with discrete boundaries or that rely largely on vector based data

Raster Models

Preferred when the data sources are largely raster based data sets

Useful if DEM or Satellite imagery is a major component

Modeling Process

Modeling Process

Define the goal of the modeling process

Isolate the factors that are likely to be important

Implement the model

Test the model to assess it’s validity

Importance of GIS in the process

GIS is a good tool for integrate a variety of spatial data

GIS can perform either raster or vector based modeling

Where necessary GIS can convert data from raster to vector or vice versa

Modeling can be done within the GIS but may also require use of other external database, statistical, graphic or analysis programs

Loose coupling requires use of external formats to transfer data

Tight coupling supplies a common user interface

Embedded system bundles the other software directly within the GIS

Model Types

Binary Models

Classifies region into binary responses (e.g. Yes/No, Present/Absent)

Index Models

Classifies region into 3 or more classes (e.g. Low, Medium, High)

Regression Models

Process Models

Model Types (cont.)

Binary Models

Binary models produce binary output

Applications

Identify areas that have changed from one state to another

Identify areas that are suitable for a purpose

Examples

Catchment area delimitation

High/Low archaeological site probability mapping

Model Types (cont.)

Index Models

Index models produce a suitability index as output

e.g. High, medium, low suitability

Weighted Linear Combinations

Create suitability criteria for each factor, then add these together in proportion to their presumed importance

Applications

Suitability Analysis

Vulnerability Analysis

Examples

Model Types (cont.)

Regression Models

Regression models compute a dependent (aka output) value based on the values of a independent (aka predictor) variables.

f(x) = a + b1x1 + b2x2 +…+ bnxn

Linear Regression Models

Produces an output value that can take on a range of values

Logistic Regression Models

Output of logistic regression is a binary response

Model Types (cont.)

Process Models

Process models using existing theories to create an analytic model

Application

Soil Erosion models