GIS as Reproducible Science
3/7/21
Because GIS is used in different ways, to different degrees, and with users and developers at different levels of involvement with the GIS community and software advancement, “doing GIS” can fall under three different categories, according to Wright et al. (1997). These categories are 1. Using GIS as a tool; 2. Advancing the capabilities and ease of use of GIS (“toolmaking”); and 3. GIS as a science. The word “science” is tricky because it can have many different definitions and implication for how GIS as a field is incorporated into academia. While not all users and developers may interact with GIS as a science, to me GIS can still be a science, given it uses the scientific method to test hypothesis using data and logic. While GIS is often seen as a subset of geography, I would argue that the two fields are separate but overlap. GIS is often applied to geography, and while spatial analyses are inherent in GIS work, it can often be applied to and require knowledge of other fields in developing software, including ecology, computer science, engineering, design, and math. GIS then becomes a new type of field, unique in that it is inherently interdisciplinary.
In this course so far, we have been thinking about GIS as a field and lens through which we may explain phenomena. I’d say that qualifies as being involved in the beginnings of “doing GIS as a science.” We have also been using GIS as a tool, and learning more about the code behind GIS which will then enable us to become toolmakers in the future; these skills are not enough to qualify as science but together with the other aspects of the course, it contributes to “GIS as science.” While GEOG0120 (Introduction to Human Geography with GIS) was mostly focused on GIS as a tool (not science), we also learned about the theories behind various spatial interactions (moving toward qualifying as science but probably not enough to qualify).
Another central aspect of this class is learning about how open source GIS works and what it has to offer to the fields of GIS, geography, and whatever other fields we can use GIS to improve, whether it be ecology, climate studies, or health care. A particular benefit of open source work is that it encourages and simplifies efforts to improve reproducibility and replicability. This is particularly important for GIS as it can be difficult to document every single decision one makes in an analysis. Because QGIS has a model maker within it, models can be made and published to the public easily, providing accessible and easy-to-understand (relatively) documentation of methods. In addition to increased accessibility to documentation, open source creates a culture and ethic of transparency, knowing others can look at one’s work. Open source models save a great deal of time in reproducing and replicating studies because one only needs to compile inputs and press “play,” rather than clicking through each step of an analysis—plus, users do not need to worry about mistakes because the model is already set. This encourages reproducibility and replicability studies; as more researchers use models, instances of checking others’ work will also increase and thus create a new norm of transparency and work-checking.
Sources
- Wright, D. J., M. F. Goodchild, and J. D. Proctor. 1997. GIS: Tool or science? Demystifying the persistent ambiguity of GIS as “tool” versus “science.” Annals of the Association of American Geographers 87 (2):346–362. DOI: 10.1111/0004-5608.872057
- National Academies of Sciences, Engineering, and Medicine. 2019. Reproducibility and Replicability in Science. Washington, D.C.: National Academies Press. Ch. 2-3. DOI: 10.17226/25303