Navigating Error and Uncertainty
Uncertainty and GIS
GIS today creates the impression that geographic representations are inherently precise, and that spatial analysis can flawlessly encapsulate our world. However, I argue that the complex tools we have today do not eliminate the errors and uncertainties that come with spatial analysis. As Longley et al. state, we become more uncertain about the quality of our digital representation with more data at our disposal. Uncertainty is an inherent aspect of geographic research, and as a GIS student, I will delve into the complexities of managing and understanding it. Today, I will explore Figure 6.1 from Longley et al.’s “Geographical Information Systems and Science” (2008) and reflect on how geographers can address uncertainty, drawing insights from Tullis and Kar’s (2021) perspective on ethical replicability and reproducibility in GIScience.
Longley et al. identify three primary sources of uncertainty in the research process, as illustrated in the figure above. It vividly illustrates the complexity of how uncertainty manifests at different stages, from initial real-world observations to subsequent laboratory analysis. The first stage involves uncertainties in the conception of the geographic phenomenon, highlighting how our initial understanding shapes the representation of it. The measurement process influences how the phenomenon can be analyzed within Geographic Information Systems (GIS). The analysis is based on an uncertain conception and measurement; hence, the phases interact with one another and lead to distortions in the representation of the real world.
My GIS project on environmental injustice in Harris County, Texas for my GEOG0120 class required me to analyze the impact of Hurricane Harvey on flooding and toxic releases in the region. This required dealing with incomplete data as we explored majority groups Asian, Black, Latinx, Mixed, and White. There was a lack of representation for the Asian majority group that experienced flooded buildings that were in the FEMA zone as there were only 3 Asian block groups whereas other block groups ranged from 175 to 864. This meant that we could not include Asian representation in our conclusion about the affected majority groups in Harris County.
Read more about my analysis in my visual essay.
Uncertainty is a constant companion in spatial research. The project presented the limitations of FEMA preliminary flood hazard data and how the uncertainty leads to more harshly affected communities. FEMA was not able to predict flooding for a large part of areas that Latinx and mixed communities currently live in which were affected the most. FEMA’s data for predicting flood in Harris County was unreliable as the predictions are based on historical flooding and development. Predictive models as such are easily outdated as the world constantly changes. The project demonstrates the complexities of real-world phenomena, especially in estimating flood risk, unpredictable climate change, and rising sea levels.
Mitigating Uncertainty
Some steps that can be taken to mitigate the problem of uncertainty are ensuring robust data collection, adopting advanced modeling techniques, and employing ethical considerations. Ensuring high-quality data collection and documenting metadata are crucial steps in reducing uncertainty at the data source. Additionally, geographers should adopt rigorous methodologies for uncertainty assessment, incorporating robust statistical techniques and sensitivity analyses to quantify and communicate uncertainty. This not only improves the reliability of research but also ensures ethical rigor, as emphasized by Tullis and Kar. Utilizing advanced modeling approaches, such as Bayesian methods or Monte Carlo simulations, can help quantify how uncertainty propagates. Moreover, geographers should actively engage in discussions about the ethical use of geographic data, advocating for responsible and unbiased approaches.
“As replicability and reproducibility (R&R) crises develop within emerging convergent inquiry, ethical use of provenance information is central to the establishment and preservation of trust in critical applications of GIScience and geospatial technologies” (Tullis and Kar, 2021).
In conclusion, data should not be regarded as absolute truth especially today when there is an abundance of data available. The data’s quality should be questioned and considered before use. Uncertainty is a ubiquitous challenge in geographic research that cannot be eliminated entirely, and geographers play a pivotal role in addressing it ethically and rigorously. By adopting transparency, robust methodologies, and ethical considerations, geographers can contribute to more reliable and responsible geographic information systems and science. This aligns with the evolving landscape of GIScience, as underscored by Tullis and Kar’s call for ethical replicability and reproducibility in critical applications.
Find more information about the class I am taking here!
Bibliography
Longley, P. A., M. F. Goodchild, D. J. Maguire, and D. W. Rhind. 2008. Geographical information systems and science 2nd ed. Chichester: Wiley. Chapter 6: Uncertainty, (pages 127-153)