David Stonham – PEI Storm Surge

Data Cleanup

Once I had downloaded my data there were several things that needed to be done to clean it up. Not least ensuring that the each layer was projected properly into the “NAD 1983 CSRS Prince Edward Island” projected coordinate system.

PEI Boundary

This shapefile had many polygons that were not technically land which I needed to manually erase from the feature class. The sea surrounding the island was one, but others included rivers crossed by a bridge close to their mouth, and enclosed harbours. Given that I was working with hydrological features (i.e., the sea) I didn’t want the river to be mistaken as part of the land mass in future processes.

DEM -GeoGratis

I was only interested in the elevation of points within the boundary of PEI (not the elevation of the surrounding sea). I couldn’t simply select all raster cells that were greater than zero, because there were many points inside the provincial boundary that are recorded as having zero elevation -equating to a high vulnerability. Eventually, I converted the PEI Boundary shapefile to a raster, assigned the points inside a value of 1, and multiplied this layer with the DEM using the raster calculator.

Buildings

The buildings layer was only available as a .dwg file, and included multiple layers. I decided to convert the only the point layer to a shapefile, and discarded the remainder of the layers which included polyline boundaries of buildings. One drawback of using this layer is that the attributes do not distinguish between a building that is used as a primary residence, and those that are commercial or outbuildings. This information would have been helpful in order to estimate the number of people affected in each building. Additionally, buildings in ~6 of the major urban centres were excluded from the file, presumably they have their own local planning departments, so these had to be excluded from my calculations.

omitted_communities — The communities of Cornwall, Charlottetown and Stratford, were among those omitted from the buildings layer.

Roads

For the most part, the road data was complete and comprehensive, I symbolised the features using the road type attribute to give an idea of the size and priority of the roads. I also used the SQL statement “ROADCLASS” <> ‘Service Lane’ to remove all of the service lanes from the data displayed.

Determining Vulnerability Zones

Using the digital elevation model (DEM) I was able to classify the data into four bands and export a shape file to represent each of the vulnerability scenarios. These shapefiles will be useful later in the analysis as I will be able to join attributes to the buildings that fall within their boundaries spatially.

At this point I also created my first map layout which shows areas of special interest. Some areas like Victoria (inset 4) will likely experience major damage in the event of an average storm event and by the end of the century, after a 1m rise in sea level, the vulnerability will be greatly exacerbated.

vulnerability_areasofinterest — Areas of particular interest regarding flood vulnerability on PEI

Vulnerable Buildings

Using the “select by location” tool I was able to assign binary attributes to the buildings falling within each of the vulnerability zones, an attribute that was necessary for further analysis of the buildings in each of the four scenarios.

buildingattributes — Additional fields in the attribute table provided tabular data useful for further analysis

Kernel Density

In order to establish areas of greater vulnerability and visually display them on my final map, I used the spatial analyst tool “kernel density” to highlight ares where there is are groups of vulnerable populations. This tool gives a good sense of the areas of importance, so I will replicate it for each of the four scenarios in the final conclusions phase.

Flow Chart

The following flowchart outlines the process I have undertaken to produce the vulnerability density map.

processchart — Flow chart describing the process undertaken to create the building vulnerability density map layout