Question

For exercise 3, I evaluated the connectivity of each building within Seaside, OR, to critical infrastructure following a rupture of the Cascadia Subduction Zone. The probability of connectivity for each building was determined using networks and considered the following:

• Electric Power Network (EPN): probability that each building has electricity.
• Transportation: probability that each building can reach the hospital or fire stations via the road network.
• Water Supply Network: probability that each building has access to running water.

The connectivity analysis was deaggregated by hazard as well as the intensity of the event.

Tool and approach

For this exercise, I used: (1) a probabilistic earthquake/tsunami damage model to evaluate the functionality of linkages; (2) the network analysis package python-igraph to evaluate the connectivity of each tax lot to critical infrastructure; and (3) QGIS for spatial visualization of the results.

Description of steps

Networks were created to represent the connectivity of the three infrastructure components (Figure 1). A network consists of nodes connected to each other through edges. When edges are removed, a connectivity analysis can be performed to determine whether there is any path from one node to any other specific node. A disconnection in the network results in two (or more) separate networks.

Here, the earthquake and tsunami hazards cause damages to edges which are removed from the network if deemed nonfunctional. A connectivity analysis between each tax lot and critical infrastructure was performed, and each tax lot was triggered with a binary yes/no for connectivity. A Monte-Carlo approach with 10,000 simulations was implemented to determine the probability of each tax lot being connected to critical infrastructure. The resulting probabilities were then added as attributes to GIS shapefiles in order to evaluate the spatial distribution of connectivity.

Figure 1: GIS and representative networks for each infrastructure component

Description of results

Characteristics of the network can be described by a degree distribution. In a network, the degree of a node is the number of immediate connections that the node has to other nodes (e.g. a node connected to 3 other nodes has a degree of 3). A histogram of the degrees can be generated to describe the overall distribution of the entire network. The degree distribution for the three infrastructure components are shown in Figure 2. It can be seen that in the EPN network, most nodes are connected to two other nodes. This is likewise apparent in the network of Figure 1, as the EPN network appears more “linear” compared to the transportation and water networks. The transportation and water networks exhibit similar characteristics to each other in that the majority of nodes have a degree of three.

Figure 2: Degree distribution for each infrastructure component

Using the results from the Monte-Carlo network analysis, maps were created to show the spatial variability of connectivity. The connectivity was deaggregated by both hazard and intensity of the event, as deaggregation provides an avenue for smart mitigation planning. Although similar maps were produced for all three networked infrastructure, for brevity, only the spatial distribution of the transportation network is shown (Figure 3). The maps show the probability of each tax lot becoming disconnected from the fire stations (2 in Seaside) and hospital (1 in Seaside) via the road network. It can be observed that the tsunami hazard results in significant damage to the transportation system relative to the earthquake hazard. The result of bridge failures caused by the tsunami can be observed for intensities larger than the 500-year event. The region west of the Necanicum River becomes completely disconnected from the fire stations and hospital which are located east of the river.

In addition to the spatial deaggregation, the aggregated results provide a comprehensive overview of the connectivity. Figure 4 shows the average fraction of tax lots disconnected from critical infrastructure across all of Seaside. The three networked infrastructure systems approach complete disconnectivity for hazard intensities larger than the 1000-year event. The transportation and water networks are dominated by the tsunami for the higher magnitude events; whereas the EPN see’s significant damage from both the tsunami and earthquake. Consequently, if resource managers are planning for high magnitude events, they should invest in tsunami damage mitigation measures.

Figure 3: Spatial distribution of connectivity to fire station and hospitals

Figure 4: Fraction of tax lots connected to critical infrastructure

Critique

Network analysis provides a means to evaluate the connectivity of tax lots to critical infrastructure, and incorporating probabilistic methods accounts for uncertainties as opposed to a deterministic approach. While this type of analysis can be useful to determine overall connectivity, it does not account for limitations and additional stresses in the “flow” of the network. For example, damage to the transportation network would result in additional travel times to the fire stations and hospital. In order to provide a more comprehensive analysis of the impact to networked infrastructure, both connectivity and flow should be considered.