Most data was acquired from the department of Geography of the University of British Colombia. These included :
- Roads layer of Greater Vancouver (2009)
- Schools in Greater Vancouver (2009)
- Libraries in Vancouver (2009)
- Community centers in Vancouver (2009)
- Hospitals in Greater Vancouver (2009)
- Landuse of Greater Vancouver (2009)
- DEM of greater Vancouver 25m resolution (2006)
- Greater Vancouver shoreline (2008)
- Census tract cartography layer of Greater Vancouver (2011)
- Basemap from Open Catalogue
Moreover, British columbia statistics open data provided information in excel format for the census tracts population census data for the year 2011.
All data was projected on a NAD_1983_UTM_Zone_10N coordinate system with a Transverse Mercator projection.
Data
Methods
Run up analysis are very common to analyze tsunamis. They consist of using elevation of the desired location and creating risk areas based on a conditional if/else evaluation on each of the input cells of an input raster.
In this case, the analysis determines 3 categories of risk: severe, high and elevated, based on elevations ranging 5-10meters, 2-5 meters and 0-5 meters.The model builder above displays the technical process. The Cons tool was used with a query for each 3 elevation. For example, in order to determine the elevation of 2 to 5 meters, the query was as such: "VALUE" >2 AND "VALUE" <=5 The three raster layer created were then combined into a single raster data layer through the Mosaic to New Raster tool. Converting this newly combined raster layer into a vector one permitted the object analysis and the precise visualization of the risk areas. The object analysis consisted of intersecting roads, schools and shelters (hospitals, libraries and community centers combined) and can is displayed here. In order to evaluate the demographic affected, I gathered information from the BC stats website through an excel sheet of the census tracts population for the year 2011. This excel document was then joined to the census tract polygon layer obtained by UBC geography to create a density map of Greater Vancouver shown below. The run up areas and the census tract were then intersected and calculations were done to understand how many people would be affected by inundation and visualize the census tracts affected map.
LEAST COST PATH
Based on the data created for the run up analysis, I then engaged in locating safe routes for specified schools in the severe risk zones to refuge in the no risk zone. 12 schools were found to be at severe risk of inundation. Bird-like flight distance was calculated from each school location to the closest refuge, creating a "partner" for each school. Although, some refuge were not the actual closest distance to a school, different ones were chosen in order to reduce the amount of crowding that could happen in times of refuge.
In order to evalute mitigation options in the event of a tsunami in the Greater Vancouver region i started by analyzing the risk area through a run up analysis. I then created least cost paths in order to facilitate the access from risky locations to safe refuge. And finally a multi criterion analysis was done in order to understand the best and worst locations for residents and officials to prepare accordingly.
RUN UP ANALYSIS
All the listed above schools and refuge centers were mapped by selecting by attribute each of their objectid and creating a new layer from the selection.
In order to create least cost path between the schools and refuge, I first had to create a cost map for different categories. This was done by working with the landuse layer, the roads layer, the run up areas layer, and a creation of buffer zones from the coastline. First water was created as a layer using the select by attribute tool from the landuse layer. Secondly the run up areas was erased from the landuse layer to create a no risk zone layer, which was then merged with the run up areas to visualize an all zones map. The roads layer was used to gather the bridges, which through select by attribute were turned into a new layer, while consecutively buffers were created around each bridge for later purposes working with rasters. Lastly, in order to create zones representing the proximity to the coast, I used the multi ring buffer to create buffers from 500meters to 8000 meters away from the coastline, shown in the map below.
I then merged the bridges and buffers layer together and the all zones layer and the water layer together respectively. The two layers created were then converted to raster with the polygon to raster tool, to ultimately add their cost together with the Plus tool which adds the values of each cell together, resulting in values from 2 to 300. The highest values representing the most inundated areas, while the lowest value the least inundated ones, such that the least cost path will try to stick as close to possible to the latter values. The output cost map is displayed here.
The model builder below shows the process effected to create the least cost path between the schools and the refuge, based on the cost map created. For each school, I used the cost distance tool which evaluates the distance away from the input point outwards. This tool gives a cost raster and backlink outputs. Those two are then used as inputs together with the refuge associated with the school used in the cost distance for the Cost Path tool. The process was thus repeated 12 times to create a path for each school in the severe risk zone. Each raster path created had to be converted to a vector polyline due to the mixed pixels problem, where raster overgeneralize the representation of the groud feature thus displaying a path that often is not even visible on the map, or doesn’t reach full destination.
After creating all the layers necessary, I then added a field to each of their attribute table, entitled “cost” to reflect the ease of traveling through such areas during a tsunami occurrence. In order to create the cost for layer that contained different categories, I had to select by attribute and then use the field calculator, such as for the “zones” layer. The higher the cost, the more difficult it is to travel through. As a result, the following table displays the ranking of such costs. For example, the severe run up area was given a cost of 100 to represent its danger, or inundated state and thus impossibility to travel through, while the “no risk zone” was attributed a cost of 1.
MULTI CRITERION EVALUATION
A multi criterion analysis was done in order to analyze suitable areas for developers, residents or officials who would like to build, reside, or develop facilities for tsunami preparedness.
The three criterions evaluated are: Elevation, proximity to coast, and density. A DEM map can be shown here. After using the Polygon to Raster tool on the elevation and the density maps, all three layers had to be normalized to a common scale 0 to 1. The Fuzzy Membership tool was used for all three layers. Values of 1 are attributed as “good” whereas 0 are considered “bad.” In the case of elevation, highest elevations were attributed a value of 1 and lowest a value of 0. In the case of proximity to coast buffer layer, small distance, such as 500meters were given a value of 0 while bigger distances were given a value of 8000meters. Finally, in the case of density, high densities were given a value of 0 while lowest densities were closer to zero. The reason for the latter factor attribution is because, in the event of tsunamis, if you are located in a high-density area, chances are that escape methods, or hospitals would be very crowded as opposed to lower density areas. Those 3 variables are displayed in the map below.
After having defined the 3 criterions and their common scale, I went onto a website called (http://mychoicemydecision.com/Default.aspx), in order to determine the weights for each factor using an analytical hierarchical approach (AHP). Based on the data found through the run analysis, the least cost path evaluation and background literature review I created weights based on the idea that elevation is more important than both density and proximity to coast, but that coast and density were of similar importance. The criteria importance created by the online company gave the following results, shown below.
By displaying all 12 paths onto a single map, I created what I hope to be the best path to undertake for endangered schools in search of refuge during or after a tsunami, considering all factors costs. Furthermore, each school could act as a meeting point for much more than simply the schools, but rather all population of that census tract and buses could drive them all to their respective refuge.
I then performed the actual weighted linear combination using the Weighted sum tool selecting the rasters created earlier with the Fuzzy Membership tool, each variables was given its associated weight as seen above. The output map can be seen below. A second map was created layering the school most at risk in order to evaluate the legitimacy of both analysis combined together and see matching correlations, and can be seen here.
Furthermore, in any MCE a sensitivity analysis should be conducted in order to compare our results to a different set of weights to understand the validity of our assumptions. Using the same tool Weighted Sum, I used the three Fuzzy Membership layer of all three variables, but this time given them all a weight of 33. This newly created layer was then itself used in the Weighted Sum tool together with our first weighted MCE output, together they created the sensitivity analysis shown below. A classification of 5 classes was done using the natural jenks method which I thought best displayed the results on a scale from best to worst.
