As we cover in various ways throughout the course, there are many aspects to spatial data and analysis that make some people say that ‘spatial is special’. Phenomenon such as spatial autocorrelation and distance decay that have consequences, such as the Modifiable Areal Unit Problem (MAUP) scale and aggregation effects, that need to be understood when working with real data. There are a number of complex concepts covered in this lecture (temporal and spatial autocorrelation, spatial sampling, distance decay), but as you will discover in your labs they do affect our interpretations of the world (as represented in a GIS, and as revealed through spatial analysis).
In these lectures I will review some of the concepts.
Learning Objectives
- Understand that Tobler's "First Law" of Geography is the formalization of spatial autocorrelation;
- Recognize the dependencies between scale and representation;
- Understand principles of building representations around geographic samples;
- Know how the properties of smoothness and continuous variation can be used to characterize geographic variation.
Required Readings
Text: Chapter 2: The Nature of Geographic Data.
Slides for the Nature of Geography Data lecture.
Alternative text: Fewer than we would like directly cover this topic.
A chapter from a book that provides a brief overview of MAUP: Wong D.W.S. (2004) The Modifiable Areal Unit Problem (MAUP). In: Janelle D.G., Warf B., Hansen K. (eds) WorldMinds: Geographical Perspectives on 100 Problems. Springer, Dordrecht.
Recommended Readings
Section 2.2 of Geospatial Analysis provides a brief overview of the nature of spatial relationships, which are the subject of this lecture.
A thorough discussion of spatial autocorrelation can be found here: CATMOG 47 Spatial Aurocorrelation by Micheal Goodchild (1986).
ESRI: Spatial Patterns of Disease Inspire New Ideas on Possible Cause
WRT Lab 1: ArcGIS Pro Resources: Optimized Hot Spot Analysis (and, FYI, Regression Analysis Basics)
An online textbook on statistics that follows the same format as the Geospatial Analysis text.
The concept of spatial heterogeneity is a difficult one to come up with relevant examples; on-the-other hand, it is easy to provide examples of spatial homogeneity. The wiki page on spatial heterogeneity illustrates the complexity of defining the term: note at the top of the page the warning that "This article has multiple issues." So, even amongst experts there are disagreements as to how to explain / define the concept. However, think about the Language diversity index you are using in Lab 1--the higher the index the more heterogeneous the area is with respect to language diversity, the lower the index the more homogeneous the area is. Also, consider that as the LDI increases (increasing heterogeneity in terms of language diversity) other factors may be becoming more homogeneous, such as income (likely more consistently lower).
Useful Resources
A Note on Intensive and Extensive Data
The Modifiable Areal Unit Problem (MAUP)
Chapter 6: GIS and small-area estimation of income, well-being and happiness in the text GIS and the Social Sciences; Theory and Applications has a helpful section on the use of multiple regression from a social science perspective (i.e., Statistical model-based estimates).
A discussion on the problem of using ZIP codes in health research (in particular, how the lead-poisoned water issues in Flint Michigan were obscured by the use of ZIP codes).
Keywords
spatial autocorrelation and the Tobler Law; scale; representation; spatial sampling; distance decay; hotspots analysis.