Evaluating Tsunami Evacuation Zones in Lincoln County, Oregon

Introduction:

Robert McCaffrey and Chris Goldfinger revealed the violent character of the Cascadia subduction zone in 1995 (McCaffrey and Goldfinger 859). Since then, the seemingly  geologically inactive coastline of the Pacific Northwest has become grave topic of concern among community members and policy makers as the unsuitable infrastructure and location has put them in a situation of impending destruction (Witter et. al 1784). Distant tsunamis have plagued the Cascadia coastline over the past century, resulting in the implementation of Tsunami hazard mitigation by state and local officials (qtd. in Oregon Tsunami Clearinghouse), however the majority of the Oregon Coast does not have evacuation systems in place to compete with the impending damage of the ‘mega-thrust’ earthquake and following Tsunami that McCaffrey, Goldfinger and others predict (Schultz). With this in mind, the following analysis focuses on evaluating the suitability of coastal land for tsunami evacuation zones, and aims to provide a overview of locations that may be used as refuge areas in Lincoln County, Oregon.

Tsunami evacuation routes must be employed so that they are compliant with the physical geography of a specific location, and so that they are as accessible to those within the inundation zone. Since the utility an tsunami evacuation map depends on multiple geologic, infrastructural and human variables, a Multi-criteria evaluation process is employed with respect to topography, distance to inundated water bodies, and accessible infrastructure. This study uses quantitative measures suggested by the literature to assess areas near the coastline and connected water bodies to determine accessible, safe areas where earthquake survivors can travel to to escape an impending tsunami.

Methods:

The general process for evaluation was to determine and define three criteria attune to the physical geography of the area that could aid in the determination of accessible, unaffected locations where resident could seek refuge. For sake of simplicity, I have included four variables that will aid in the determination of refuge areas: distance from inundated water bodies, elevation, distance from roads, and cost of travel from inundation zones with respect to slope. The definition of these variables are elaborated on below:

The ‘distance from inundated water bodies’ was represented as a range of distance beginning at the affected bodies of water to 3200 meters – the estimated distance an individual could travel on foot – amidst damaged infrastructure – after an earthquake. This space is identified as our primary area of concern; all other data sets included in the study span the 0 – 3200 meter area, since areas outside of this space are considered ‘unreachable’ to an individual inside the inundation zone.

Elevation is included in the study as a binary dataset that classifies ‘safe’ areas as having a value of 1, and ‘unsafe’ areas having a value of 0. Using a worst-case approach to classifying safe-zones (the philosophy and standard of Oregon planners (Oregon Tsunami Clearinghouse)), I set the distinction at 30 meters –  the largest predicted surge from the tsunami (Witter et. al 1801).

Since it is desirable to place tsunami evacuation zones near road networks where they will be accessible and easy to navigate by local residents, the distance from roads becomes a necessary component. The distance an individual is expected to travel off of a roads has been classified as 400 meters to account for alternate land cover in the area (Wood et. al).

High slope in an area may be desirable as it implies a higher elevation is reached in less of a horizontal travel distance, on the other hand, an area of high slope (especially one that is off of a road) could indicate impassible terrain. To highlight beneficial terrain, a cost-distance surface was employed to indicate the overall relief of the landscape with respect to the inundation zone, and restricted to areas near roads.

After defining the variables above, each data set received a standardized ‘suitability score’ ranging from 0 to 1. Once each data set was standardized, numerical weights could be applied to each data set to form a model where the weights would indicate the influence of a variable on each cell’s suitability rating. The diagram below shows the generic construction of a tsunami evacuation zone evaluation model.

Results:

Following this methodology, two contrasting models have been constructed.

The first model constructed is based on a ‘worst-case’ analysis. The objective of this model is to determine areas that are out of an exaggerated inundation zone. In the case of the most powerful shaking, it is reasonable to assume much of the road network will be damaged (Wood et. al), and so roads are de-prioritized as an indication of accessibility. A weight of .4 was given to the elevation data set, while weights of .25, .25 and .1 were applied to slope, distance from water bodies, and roads, respectively.

This model resulted in a clear distinction between areas within and outside of the inundation zone with respect to elevation and road networks. Roads near inundated water bodies are given a dramatically lower suitability score than those inland, at higher areas of elevation.

A case-study of Siletz Bay exemplifies this more clearly; the steep contrast in values show that inland areas of high relief are considered highly suitable evacuation zones.

The model above was contrasted with an ‘accessibility model’ which focuses on highlighting more accessible safety points at the expense of predicted wave-height. This is reconciled as Witter et. al describes the worst-case scenario accounting for only 5-10% of the predicted wave-height scenarios that could occur after the earthquake (Witter et. al 1804). In this case, priority was given to the road networks, while elevation and distance from the water bodies were de-prioritized, resulting in the weights .4, .3, .15 and .15 to  distance to roads, elevation, slope, and distance from water bodies, respectively.

This model provides a much more liberal view of evacuation zones, as it shows less contrast between areas inside of and above the worst-case inundation zone, as expected.

In this case, the roads in downtown Lincoln City become much more defined, as they gain a higher suitability score. Meanwhile, the areas in the previous model that had been defined as highly suitable areas do not contrast as heavily with the surrounding area.

It should be noted that I refrain from choosing specific locations to designate as evacuation routes or zones; the focus of this analysis is not to designate a specific location, but to display suitable, and readily accessible areas for any resident along the coast of Lincoln County to access in the case of a large earthquake and tsunami.

Summary: 

In light of the literature, this study has provided a much-needed overview of accessible areas outside of the expected inundation zone through a multi-criteria evaluation analysis that is attune to the physical geography of Lincoln County, Oregon. With respect to the analysis performed above, it is seen that suitable, accessible tsunami evacuation areas are determined through unique combinations of spatial variables. When elevation is prioritized over other variables, the areas deemed as ‘suitable’ evacuation areas within the study area become much more sparse. By contrast, when elevation is de-prioritized, and accessible infrastructure and topography is prioritized, we highlight more possible refuge areas.

The different models exemplified sensitivities in this model by evaluating the area based on different weighting parameters. Particularly, the polarization of safe and unsafe elevations became less apparent when the binary elevation model was associated with a lower weight in the Accessibility model. Additionally, as road networks became less weighted and elevation received a higher weight, the majority of tsunami evacuation areas appeared inland, where fewer roads exist and where elevation is typically higher. This also encouraged the binary elevation model’s influence, by drawing weight away from the roads in low-lying urban areas.

References

Mccaffrey, R., and C. Goldfinger. “Forearc Deformation and Great Subduction Earthquakes: Implications for Cascadia Offshore Earthquake Potential.” Science 267.5199 (1995): 856-59. ProQuest. Web. 14 Nov.

“Oregon Tsunami Clearinghouse.” Oregon Tsunami Information Clearinghouse. Oregon Geology, 29 Apr. 2015. Web. 17 Nov. 2015.

Schultz, Kathryn. “The Earthquake That Will Devastate the Pacific Northwest.” The New Yorker. The New Yorker, 20 July. 2015. Web. 17 Nov. 2015.

Witter, R. C., Y. J. Zhang, K. Wang, G. R. Priest, C. Goldfinger, L. Stimely, J. T. English, and P. A. Ferro. “Simulated Tsunami Inundation for a Range of Cascadia Megathrust Earthquake Scenarios at Bandon, Oregon, USA.” Geosphere 9.6 (2013): 1783-803. EBSCOhost. Web. 14 Nov. 2015.

Wood, Nathan J., Jeanne Jones, Seth Spielman, and Mathew C. Schmidtlein. “Community Clusters of Tsunami Vulnerability in the US Pacific Northwest.” Proceedings of the National Academy of Sciences Proc Natl Acad Sci USA 112.17 (2015): 5354-359. Proceedings of the National Academy of Sciences. Web. 14 Nov. 2015.