Uncertainty and Error in Geography
One of the things that drew me the most to Geography was its blend of science and humanities, making it a unique field that investigates phenomena from a multitude of perspectives. However, this blend of quantitative and qualitative elements to the field tends to cause uncertainty in many studies. As discussed by Longley et al.’s piece on error and uncertainty, the nature of the field — understanding the world — lends itself to uncertain conclusions. Because all representations of the world are incomplete, it is almost impossible to accurately portray the world’s various attributes on a map.
Maps are often taken as fact; a truth that goes unquestioned. They have an immense capability to control and alter how people perceive the world, and whose stories are told. However, they reflect the cartographer’s view of reality. They are a subjective representation of a landscape that often reveals only one perspective. Mapmaking is often considered quantitative, however, many things that Geographers seek to map aren’t bounded by rigid lines. These things like animal habitat, migration patterns, or the demographic makeup of a city all require decisions that are not arbitrary. Through this arises uncertainty and potential for error in the way the world and its spatial patterns are conveyed.
I, like I imagine many others, previously took maps at face value. But through mapping amorphous phenomena and creating classification schemes in Google Earth Engine, I’ve realized just how subjective Geography can be, even in a field that seems so technical and objective like remote sensing. While satellite imagery is an effective way to create land classifications and its analysis produces very valuable knowledge, nothing will be as accurate as on-ground classifications, which is not always possible. Further, oftentimes one area is made up of multiple classifications which makes it difficult to categorize in analysis. I’ve struggled with this when trying to map the racial makeup of a city — what most accurately reflects the ground truth makeup of an area? A choropleth map showing the majority? A dot map?
Longley et al. explain “uncertainty” to be an umbrella term to describe problems that arise out of imperfections of GIS analyses. It is inevitable that due to the nature of GIS’s capabilities and available data sources that uncertainty and error will arise. Longley et al. describe various ways in which we can mitigate error or evaluate the extent of the error, through things like confusion matrices, cross-referencing data, and spatial autocorrelation. As Geographers, it is first and foremost important to acknowledge the uncertainty in our work and be inclusive about our decisions. Geography, like many other sciences, stem from a white, Eurocentric perspective which unfortunately has persisted today and manifests itself through subconscious (or conscious) analytical or visual decisions. Marginalized voices and narratives are often silenced, seen through the lack of Indigenous knowledge representation. Thus, while it’s important to employ the methods Longley et al. laid out about addressing uncertainty, we must go beyond technical methods and acknowledge the faults in inclusivity that deny validity to some types of knowledge and prioritize “scientific,” published knowledge.
Sources:
NASEM. 2019. Reproducibility and Replicability in Science. Washington, D.C.: National Academies Press. https://www.nap.edu/catalog/25303.
Longley, P. A., M. F. Goodchild, D. J. Maguire, and D. W. Rhind. 2008. Geographical information systems and science 2nd ed. Chichester: Wiley. (only chapter 6: Uncertainty, pages 127-153)