In a paper for An International Journal, Volker C. Radeloff et al. (2000) look to demonstrate the value of GIS through examining the spatial pattern of housing density in the northwest Pine Barrens region in Wisconsin and its relationship to nearby lakes and land-cover. The authors briefly review the historical troubles spatial geographers had with research that was confined to disciplinary boundaries and highlight the ability for GIS to integrate different kinds of data to create a sense of interdisciplinary work (p.600). The primary argument that Radeloff et al. present is that “the underlying causal relationships between housing density and land cover probably work in both directions” (p. 604). In the paper, the authors were keen to note that the purpose of study was to identify associations between the two variables. Due to the nature of their analysis, results produced revelations about correlation, but not causation. Between the two variables, there is not a set dependent or independent variable.
Methods / Procedures:
Housing unit density was calculated through data taken from the US Census Bureau 1990 Census of Population and Housing. Radeloff et al. collected total housing units and land area at the census block level. Land-cover data was classified using multi-temporal Landsat imagery at a resolution of 28.5m. 30 identified land-cover classes were grouped into 8 classes for the study: pine forest, other conifer, oak, other deciduous, brush, grass, water, and artificial. Two different methods were used to integrate census data and land-cover classification. GIS was used to overlay census data on top of land-cover classification to qualitatively examine spatial patterns (p.601–602). Second, Radeloff et al. calculated the relative area of land-cover classes within each housing density class and vice-versa to make a quantitative assessment. Housing density classes were formed by separating census blocks into 7 categories based on their density values.
For the purposes of this study, the two variables examined were appropriate, and the utilisation of two different techniques allowed for an effective analysis of the data that was collected. Having said that, the breadth of variables that were considered seemed to be slightly lacking. While housing density and land-cover may have quantifiable effects on each other, there are many other factors that affect each variable. In order to have provided a clearer image of the different relationships, it may have been more effective to also include factors such as mean household income and more comprehensive land-use data.
Evidence:
Through their analysis, the authors were able to find evidence that land-cover and housing density were strongly correlated. They highlighted three examples and introduced relevant hypotheses. Relative abundance of water increases with increasing housing density. The authors suggest this to be an example where land-cover type influences housing density because “people prefer to site their homes along the shorelines of lakes and rivers” (p.604). Grassy land cover also presents a strong relationship, but one that suggests land-use (i.e. housing) influences land cover. The theory here is that herbaceous cover hosts a large share of medium-housing-density area due to prominent agricultural activity (605). Relative abundance of pine reveals two levels of housing density where pine presence peaks. This is an example of multiple interactions between land cover and housing density. Pine plantations equate to a high abundance of pine in low housing density areas, and pine is also abundant in areas of high housing density where the soils were unsuitable to farming and land was instead used for other purposes.
Validity:
Based on the data provided and analysis that was carried out, Radeloff et al. present a strong case that land-cover and housing density may be strictly correlated. They take care to clarify the study was purely to indicate correlation, and that determining causality requires further study. Based on the broad nature of this study, it is correct to state that the authors’ claims here are valid and strongly supported by the evidence that they present. Because this study seemed to have been conducted with the combined purpose of advertising the ability of modern GIS technologies as well as the stated purpose, I would rate this paper a 8.5/10. Procedures followed were clear in their purpose and achieved their stated goal. The study contained nothing that was especially extraordinary and leave plenty of opportunity for further research.
Source:
Volker C. Radeloff, Alice E. Hagen, Paul R. Voss, Donald R. Field, David J. Mladenoff (2000) Exploring the Spatial Relationship Between Census and Land-Cover Data. Society & Natural Resources, 13(6), 599-609, DOI: 10.1080/08941920050114646