An Example of Sensitivity Analysis

These materials may be used for study, research, and education, but please credit the authors and source: Donald J. Huebner and Kenneth E. Foote, The Geographer's Craft Project, Department of Geography, University of Texas at Austin. All commercial rights reserved. Copyright 1996 Donald J. Huebner and Kenneth E. Foote. Please send your suggestions or comments to k.foote@mail.utexas.edu or djhuebner@mail.utexas.edu.


Analyzing a Water Supply.

A small town adjacent to both a national forest and an air force base must increase its water capacity. The city hired a consulting firm to assist water board planners in determining different courses of action to increase municipal water capacity. Using GIS analysis based on geologic, hydrological, land use data, and proximity to the town, the consultant determined four well sites are suitable to meet the town's needs. Although each site is suitable there are several options that must be considered before choosing the final site.

Water from the wells can be piped via the shortest route or by using existing rights-of-way (ROW). The cost is variable due to distance and trenching difficulty. Water may also be treated either on site or piped raw to the current city treatment plant. For the purposes of this example, drilling costs are constant. Therefore, each site has four variable costs depending on piping route and location of treatment.

There is no best solution. Political or policy considerations may require a solution that is not necessarily the least expensive, in other words, cost may not be the only factor in the decision-making process. Instead each site is ranked according to the variables. Each well site and its variables are examined below.

As you can see in the following table, none of the options are the optimal solution for each case. Also increasing the number of variables, such as different drilling costs, quality of water, allowances for unknown factors, and production life would increase the number of permutations and further complicates site ranking. Additionally, if variable are changed for a site, for example, new or different data becomes available, the ranking will probably also change. In other words, the answer is not always "cut and dried" for a solution.

In this case each option has advantages and disadvantages and is ranked accordingly. A high rank for one option may be offset by a lower ranking for another.


Return to previous topic

Return to Managing Error


Created 25 February 1996 DJH. Revised on 18 October 1999. LNC