Open Source GIS and Reproducibility

Across scientific disciplines, including geography, the reproducibility of experiments can prove challenging. Here, reproducibility means a secondary researcher recomputing an original researcher’s study with access to their code, data, and methods. ¹ The widespread adaptation of computational analysis within science has increased the potential error within experiments. A single line of code or misused program can greatly modify results without those running the analysis’s knowledge. Updates, discontinuation, and the price barriers of some software can decrease the reproducibility of an experiment or analysis. While updates and discontinuation impact open source software, the black box nature of proprietary software can make reproducibly over a timescale less completable. For most proprietary geospatial software, source code is not available or is strictly copyrighted. Without access to the original source code, it can be challenging or impossible for researchers reproducing studies to document where their process went astray from the original.

Open source geospatial software or geographic information systems (GIS) can improve a researchers ability to replicate a study. With access to original source code, a researcher can implement the same computational tools as the original study. If the reproduced results do not match the originals, a researcher can analysis their code and the original code to make sure it would produce the same results. This is most beneficial when an experiments method, including code, are properly documented. Open source GIS also enables the adaptation of methods originally employed in the study to be applied in new situations. Open source GIS software being free to use and further develop lowers the economic boundary for a study to be reproduced. Both open source GIS and reproducibility in science share similar principles; knowledge should not be gatekept. Combining the scientific principle of reproducibility with open source GIS contributes to the development of both.

Open source GIS does not solve all geography’s reproducibility issues.² Like proprietary software, the base code of open source programs can change. While older editions of open source base code maybe possible to find, it still proves a barrier to reproducing studies. Bad data creates bad result; a GIS being open source does not change this. Further, open source GIS can aid methods documentation; however, it still rest upon the researcher to properly document their procedure. The idea of open source can be contrary to existing views of reliability and accuracy. There can be a biased towards products developed by private industry since people may believe that profit motivates private companies to produce better programs. An open source approach can be viewed as amateur or unprofessional.

Open source GIS does not solve all of geographies reproducibility challenges; however, it decreases them. Further development and adaption of open source GIS in geography can directly compliment the advancement of reproducibility in geography.

---

Sources

1. NASEM. 2019. “Reproducibility and Replicability in Science.” Washington, D.C.: National Academies Press. DOI: 10.17226/25303 ↩

2. Rey, S. J. 2009. “Show Me the Code: Spatial Analysis and Open Source.” Journal of Geographical Systems 11 (2):191–207. DOI 10.1007/s10109-009-0086-8 ↩