Vt Mobile Home Flooding Reproduction
Reproducing a QGIS Spatial Analysis of Flood Hazard Vulnerability in Vermont’s Mobile Homes Using a Code-based Approach in R
In Middlebury College’s Human Geography with GIS course (GEOG 0261), one of the major assignments for students to complete is a spatial analysis on Flood Hazard Vulnerability in Vermont’s Mobile Homes. The analysis performed by the students in GEOG 0261 seeks to identify how many mobile homes are at risk of flooding in the four southernmost counties in Vermont (Bennington, Rutland, Windsor, Windham) and compare how two methods/approaches for identifying flood risk - FEMA 100 year Flood Zones and Vermont River Corridors (from Flood Ready Vermont) vary in their identification of mobile homes at risk, where E911 data is used to locate mobile home points. The analysis is loosely based on the methodology of Baker et al. (2014), who investigated mobile home flood risk in the wake of the devastating flooding in Vermont during Hurricane Irene. The course is taught using QGIS, and all students execute the Flood Hazard Vulnerability in Vermont’s Mobile Homes using the QGIS GUI, following a specific workflow.
In the name of furthering reproducible and open-source research practices, I decided to recreate the assignment/analysis/study using a coding-based GIS approach in R. I wanted to see if results of the study are in identifying mobile home flood risk, regardless of GIS program used and regardles of whether you use a GUI or code approach. Because the QGIS approach relies on clicking a lot of very specific buttons in a very specific order, it is easy to make mistakes and accidentially deviate from the intended workflow. It is also difficult to know/see a record of the changes that you have made to different layers and attribute tables.
The code-based approach in R uses a dedicated codebook, which contains all the code necessary to complete the entire analysis all in one place. It can be run all at once to produce the results. I find that the code-based approach in R perfectly recreates the results produced using the QGIS workflow (except for a slight variation due to a different shapefile of town polygons that I used…see the report for discussion on this). This lends support to potentially teaching students of either GEOG 0261, or students in higher level GIS courses, a code-based approach to spatial analysis. The code-based approach, although more difficult to visualize, results in greater likelihood of accidential error when executing the workflow.
Additionally, I sought to reduce uncertainty/threats to internal validity in the analysis results by addressing a boundary distortion that occurs in the original study along the eastern border of the state, at the Connecticut River. I find that the omission of a River Corridor along the Connecticut River leads to several hundred mobile homes not being identified as at-risk of flooding when using the River Corridor approach.
View my report HERE
View my repository for this reproduction study HERE
References:
Baker, D., Hamshaw, S. D., & Hamshaw, K. A. (2014). Rapid flood exposure assessment of Vermont mobile home parks following Tropical Storm Irene. Natural Hazards Review, 15(1), 27-37. DOI: 10.1061/(ASCE)NH.1527-6996.0000112.