Discussion: Map 1

The first map suggests that the CT centroids that are located closer to the safety hubs are better allocated (with respect to the cutoff/impedance value) than those that are further away. This could mean that the people who are closest to the safety hubs may actually approach them in the case of a natural disaster, given their proximity to them. However, people who are further away from them, may consider to just remain in their home (should their home withstand the magnitude of the earthquake) because moving through the city after the natural disaster has struck may not be the best idea unless help is really needed. Additionally, choosing an impedance cutoff was a challenge because the City of Vancouver website does not provide any information regarding the capacity of each hub, so we simply calculated a value that seemed logical and used the smallest value of the green CT population and the highest of the light brown CT population (because these are the two most abundant colours ranges seen in the city. Furthermore, we concluded that 6500 people was a manageable amount of people for these disaster hubs at once. However, we do not think that 6500 people or more would approach the safety hub all at once, but rather throughout the day of and/ or following the natural disaster, and this is something that cannot be taken into consideration with the Network Analyst toolset.

Discussion: Map 2

Assuming a worst-case scenario situation for Map 2, the 7 hubs that were in danger or near dangerous areas, were removed, allowing us to run the tool again. This showed a very uneven distribution in the allocation of the CT centroids to each hub. We see that when these 7 hubs are removed, there is quite a bit of pressure placed on the remaining hubs. The greatest strain appears to be placed on the hubs in the West End, where 1 hub is expected to support roughly 12 CT centroids and their populations. A similar case is seen in the East Vancouver area. Furthermore, since the support hubs are presenting ranges greatly (from 1 up to 10 or 12), we notice that there is a rather uneven distribution of support centres by population. Interestingly, regardless of the removal of the 7 hubs, the Hastings disaster hub is still ignored by the program, meaning there are a few areas of error in the tool.

However, these new arrangements showed a wider range in the percentage of impedance allocated from each CT, meaning that there are CT centroids so far away from their designated hubs that it would be more logical to stay put than to attempt moving to a safety hub. Indeed, it is unrealistic that one safety hub would accommodate people from 10 or 12 CTs–as seen in the worst cases. Despite the impedance value being set at 6500, the accumulation of people in seek of help from that many CTs would likely surpass the capacity of just one hub.

A source of error for this part of the analysis (applies to both Map 1 and Map 2) is that the tool must allocate an impedance percentage for each of the CT centroids. In other words, for each CT centroid a percentage must be allocated (in this case all CT centroids were assigned a value of 6500) . However, since each hub allocates more than one CT, the number of people seeking help in certain safety hub would greatly surpass 6500 people. If each hub were allocating just one CT centroid (ie: one hub for each CT centroid), this would not be an issue, but allocating multiple CT centroids in one hub may prove to be a problem.

The Network Analyst toolset works in such a way that it does not take into consideration the cummulative effect of people from each CT gathering per safety hub. This is a source of error that is seen throughout our analysis, particularly in Maps 1 and 2. Perhaps a way in which this could be solved is by lowering the impedance value, so that the cumulative percentages add up to a more realistic and manageable number for each disaster hub.

Discussion: Maps 3 & 4

On this part of the analysis, the tool selected more disaster facilities, better balancing the distribution of hubs within the city. Once analyzing the map of Vancouver with these facilities suggested as potential safety hub, the distribution of people per safety hub considerably improves. At this point, ArcMap suggested a total of 73 facilities between current and potential safety hubs. However, the program excluded some facilities due to their location as they appeared to be too clustered in some sections causing a redundancy in facilities in certain areas. On this note, ArcMap utilized 50 out of 73 available facilities still yielding optimal results. Indeed, in most cases the projected after-earthquake maximum capacity per safety hub decreased from 98 – 99% to 60 – 75%, with only 3 hubs falling under 40% of projected usage. 

In real life, having more places available in the case of a natural disaster would still make a lot of sense, because the city is quite densely populated. For this matter, we have kept all facilities (used in the analysis shown in Map 3) in Map 4 because even if the tool does not consider them, we would like to leave the option open to other, unexpected, factors in an actual megathrust earthquake that might determine if they will be used or not.

Additionally, some areas, like Point Grey and Kitsilano, still show a fairly uneven distribution of many CT centroids moving towards one hub or one school. This, once again, is a function of the fact that the tool is trying to maximize attendance and some of these centroids are rather far from the hub/school, meaning that they are likely to accommodate a smaller percentage of the impedance.

Furthermore, there are a few errors to account for. The main one being that some of these new locations/current hubs are accommodating centroids that, literally, cross water. The tool is not aware that there are areas it cannot cross over. This did not happen with less “proposed facilities” in the first two maps, however, when there are too many, the tool attempts to allocate the centroids to the closest facility possible. This, however, only occurs in one instance.

Finally, in the case of a disaster, the new hubs would better serve the city because they are more spread out, serving a maximum of 6500 people each. Most serving ~95% of the 6500 people. In the west end, all community centres and schools are used – there is a great need for this area to have more disaster support due to poor infrastructure and high population densities as well as lower incomes in some areas.

 

Sources of Error & uncertainty

Some potential errors include: 

  • Road network update – given the constant renovation and updates to the city’s network, our analysis is subject to become obsolete in the near future if the map is not dilligently kept up to date.
  • The surficial geology layer that was obtained contains general information about Vancouver’s underlying soil types which could be improved in detail and description as its resolution is coarse.
  • Liquefaction in unpredicted areas of the city. Soil liquefaction in the event of an earthquake could compromise the city’s street network in ways that are not predicted in the map. Even though at-risk areas were avoided as much as possible, unexpected impacts to the city’s mobility paths may arise.
  • When finding schools as new venues for safety hubs, no information about their size could be found. This might limit the actual suitability of schools to accommodate people after an earthquake.
  • Population cutoff of 6500 for CT centroids during Network analysis was deliberately chosen. Impedence should be dependent on actual size and preparedness of facilities.
  • The network analysis crossed water at one point in English Bay and despite this error being corrected, all results should be carefully evaluated to make sure the network analysis is correct.

Areas of Further Analysis

  • Further analysis on suitability of chosen safety hubs should be done. This analysis should include size, capacity and general conditions of safety hubs.