User:Geo arbo/GES 679/assignment 03

Summary of Martin's Chapter 4, Theories of GIS

In chapter 4, Theories of GIS David Martin observed that while much of the theoretical work on Geographic Information Systems (GIS) explains components of a GIS and the operations that can be performed adequately, the issues of spatial data handling and spatial data processing were not well covered. The process of map making involves modeling objects that exist in the world geographic space and processing raw data into a representation. Martin described data transformation processes and suggested best practices for effectively representing spatial data.

A transformation-based view of GIS operation, Adapted from Figure 4.4 Martin(1995)

Martin observed that, in socioeconomic data analyses, classifications of potential GIS operations were plentiful but discussions of how data should be modeled were scant. He then gave an overview of GIS theory and discussed best practices for representing GIS spatial objects. Martin explained how data are processed in a GIS and described four stages of data transformation. T₁ data collection, T₂ data input, T₃ data manipulation and T₄ data output. Transformations are important because they control the evolution of descriptions of real world phenomena to geospatial analyses and map images. Martin noted that while use of GIS data transformations make it possible to modify study data to reach digital mapping theoretical goals, it “is equally possible to mishandle or unintentionally distort the digital map at this stage”.

Martin noted that socioeconomic data, such as census data, are collected, aggregated and processed as areal units, and that the analyses could be affected by the Ecological Fallacy and the Modifiable Area Unit Problem (MAUP). He said that socioeconomic data were sometimes presented without a discussion of how the data were modeled and made the assumption that the data depicted accurately represented real world phenomena. Martin emphasized the need for special treatment of socioeconomic data.

Data capture and preparation introduces the MAUP in geospatial analyses in a number of ways. Due to temporal and financial limitations, spatial raw data collections (T₁) are usually subsets of the real world phenomena being studied and therefore do not contain ‘all’ the data. The same time and money constraints can affect the geospatial accuracy of the locations of raw data points. Also, whenever raw data are projected, there is some distortion.

Data preparation contributes to MAUP issues during reclassification (T₃) operations. Martin gave an example of how the practice of aggregation of census data collected about individuals and then representation as areal data can lead to issues when it is interpreted because point data have been reclassified as areal data. Individual data are lost in the aggregation process. While a few individual data values may happen to have the same value as the aggregated arithmetic mean zone value, outlying values will not be represented. Another concern was that outlying individual values could skew data averages and that the majority of the data could be misrepresented when set members were aggregated.

Martin argued that the likelihood of unintentional error introduction at each transformation step can result in an accumulation of uncertainty that can have significant effects on socioeconomic analytical results.

References

• Martin. D. (1995). Theories of GIS. In Geographic Information Systems and their Socioeconomic Applications. New York, Routledge; pp. 44 – 64