Geoprocessing and Decision Support

Introduction

One of the most important aspects while using geotechnologies is the GIS potential to extract new information from a geographical database. This capacity is fundamental for applications as territorial ordering and environmental impact studies, where the final information has to be deducted and compiled from a basic survey. It is also relevant in socioeconomic studies, when it is required to establish indicators that allow a quantitative vision of the spatial information.

What is the new information production big challenge in a GIS? The capability to compare and evaluate the different possibilities for generating new maps. As the GIS offers a large quantity of Maps Algebra functions, it is not always easy to choose the most adequate way for data combination, related to an application.

In this context, it is useful to have decision supporting tools, that helps organizing and establishing a data combination rational model. The SPRING has a decision supporting tool for Geoprocessing based on the AHP technique (Analytical Hierarchic Process).


Decision Support - Basic Concepts

To decide is to choose among alternatives. Based on this definition the data manipulation process in a geographical information system can be viewed as a way to generate different hypothesis about a studying topic.

The fundamental concept of several decision making models is rationality. According to this principle, people and organizations follow a selection behavior, based on judgment objective criteria, based on a pre-established level of expectation.

The rational model for decision making involves four steps that have to be followed for an appropriate selection:

  • Problem definition: formulate the problem as if there is a requirement to get into a new state..
  • Search for alternatives: check for the possible solutions alternatives and determine an evaluation criterion.
  • Evaluate the possibilities: each solution alternative should be evaluated.
  • Alternative selection: the possible solutions are ordered, the idea is to select the best evaluated alternative or to select the best ones for further evaluation.

The AHP Technique - Analytical Hierarchic Process

When there are different factors contributing for the decision, how to define the relative contribution of each factor? In order to solve this problem, Thomas Saaty proposed, in 1978, a selection technique based on a logical comparison of pairs. In this procedure, the different factors influencing the decision process are compared pair-by-pair, and a relative decision criterion is attributed to the pair relationship, following a pre defined scale (see table).

AHP Scale Values for Comparison pair-by-pair
Importance Intensity Definition and Explanation
1 Equal importance - both factors contributing equally for the goal
3 Moderate Importance - one factor is slightly more important than the other
5 Essential Importance - one factor is clearly more important than the other
7 Demonstrated Importance - one factor is stronger than the other and its relevance is demonstrated in practical situations
9 Extreme Importance - The evidence differentiating the factors has the highest possible order.
2,4,6,8 Intermediate values among judgments - possibility for additional compromises

After establishing the comparison criteria for each factor combination it is possible to determine an optimal set of weights that can be used for the combination of different maps.


Decision Support System in Geoprocessing

Consider one of the most common situation in a GIS: area classification. This problem occurs in a large number of applications, such as mineral prospection, area selection for a commercial application, zoning.

For instance, consider the case of environmental preservation study in hillside areas, to establish an occupation policy, associated to environmental and creep risks maps. In order to get this suppose there is a topographical map, from a geotechnique chart, and a usage and soil occupation map (obtained from a photo-interpretation or classification of satellite digital images).

The traditional analysis procedure is based on the “same order spatial intersection sets” principle (Yves Lacoste) and it is based on conditioning (“the maximum risk occurs in areas where declivity is greater than 10%, they are not environmental preservation areas, and the terrain type is inadequate”). This analogical methodology transposition for the GIS environment needs Boolean operations usage (OR, AND, NOT) to express different conditions. This technique uses the computer as a drawing automatic tool, ignoring all the numeric processing potential in the GIS, and generates discontinuities not present in the original data. For instance, areas with declivity equals 9,9% will be classified differently then regions with inclination of 10,1%, not considering other conditions.

Maps are data and not just a drawing. Treating maps as data means to give it a numerical value for each location. This value represents a measure about something being studied, in most cases this data is stored using the raster format, which is more adequate to represent a continuous spatial data.

In the case being considered the spatial analysis using a GIS will be better when using a continuous classification technique: the data is transformed to the reference space [0..1] and processed by numerical combination, using a weighted average or fuzzy inference. Instead of a thematic map with rigid boundaries generated by Boolean operations, the result is a decision surface represented as a numerical grid. What does this result mean? A continuous view of how the measures change (such as mineralization indicator, environment susceptibility, or plantation proportions).

In the example mentioned, the result will be a numerical grid indicating, for each location, the collapsing risk, as a value between 0% and 100%. What is the advantage of this situation? It allows to build scenarios (for instance a risk of 10% or 20%), indicating different situations for decision making (emphasizing the environment protection or minimizing economical costs). In this way we get more flexibility and a better understanding about spatial problems.

Decision Supporting in the SPRING

In SPRING, it is possible to use the AHP decision technique to establish an optimal alternatives combination. Thus, follow the steps below.


Executing a decision support analysis:

  • activate the database and the Project that has the data model definitions;
  • click on Spatial Analysis - Decision Assistance (AHP)...;
  • in the Assistance for Decision window the Categories list presents only the categories associated to the thematic, numerical or image models in the database. Select at least 2 and at most 5 categories. If more than five categories are selected, a message will be displayed informing that only the first five will be considered;
  • click on View. Notice that the categories (a comparison among the different criteria), two by two, will be presented in the fields below.
  • select for each category pair the desired Weight. Notice that the corresponding values are presented to the left of each button;
  • click on the <=> button if you want to swap the order between each category pair;
  • notice that the Consistency Reason value is computed every time the weight is changed. If the value is greater than 0.1 a message will be displayed before computing the weights in the new created program;
  • click on Calculate Weight. The Save As window will be presented so the user can select the directory where the program will be stored. The program will be saved using the LEGAL language.

NOTE: After selecting which factors to combine and establishing the relative importance for each one of them, the system will give a consistency indication for the judgment (indicated in the "consistency ratio" item). Experts in AHP mentioned that it is desirable to have a consistency index always less than 0.1. If the value is greater than 0.1 one should consider to remake the judgment.

As a result, this SPRING function generates a program skeleton in LEGAL. This skeleton should be filled by the user with specific information about the data which will be used in the procedure. Remember that the AHP technique considers this weighted average. Thus, the data has to be converted to a [0..1] scale before using the program.

See also:
Programming using LEGAL
SPRING Analysis Options
Geostatistics
Image Processing