How much typhoon damage the Visayas transport network can take
- Article
- Network Science
The Philippines absorbs a large share of the world’s disaster damage. Between 2000 and 2016 the country recorded 565 disasters, Php 62.4 billion in damages, and 125 million affected families. After each one, relief goods have to move over roads and through ports that the same disaster may have blocked. The Visayas, a group of islands with many far-flung towns, is the hardest case.
This post, published by UP Data Science Society on April 12, 2024 with content by Dhan Mikhail Perdon and layout by Franczine Tan, summarizes a 2023 paper by J.C. Albert C. Peralta, May T. Lim, and Reinabelle C. Reyes in the Philippine Journal of Science.
Method
The researchers modeled the Visayas transport network as a graph. Nodes are places (municipalities, cities, seaports, and food hubs) and edges are the land and sea routes between them. Road data came from the Humanitarian Data Exchange; sea routes came from HotOSM, Google Maps, and the Visayas General Logistics Planning Map. They computed standard network measures (betweenness centrality, closeness centrality, clustering coefficient) and the Malloy-Reed criterion, a stability score where a value above 2 means the network holds together after damage.
To simulate a typhoon, they removed nodes and edges along the recorded path of Bagyong Yolanda (Typhoon Haiyan) from the JWTC best-track data. For each damage level they used Dijkstra’s algorithm to find the shortest route from a food hub to every town center, then compared the damaged travel time with the baseline to get an efficiency score between 0 and 1.
Findings
The undamaged network is stable, with a Malloy-Reed parameter of 2.98. With no damage, half of the Visayas can be reached from a food hub within 3.5 hours and three quarters within 5.5 hours, but the last quarter takes 13 hours.
Damage bites quickly. At 0.5 percent of nodes removed, the farthest towns see an extra 4 to 12 hours of delay. At 1 percent, 30 of the 251 town centers are cut off entirely and the rest wait two to three times as long. At 5 percent, almost every town beyond a 10 kilometer radius of the food hub is disconnected. Efficiency stays above 0.8 while damage is under 1 percent, drops sharply at 3 percent (about four times the normal travel time), and falls below 0.2 beyond that, which means travel times more than eight times normal.
What the authors recommend
Because distant towns suffer even at small damage levels, the authors argue for a decentralized approach: preposition relief goods closer to where they will be needed and diversify sea routes so islands have more than one way in. The paper includes a prototype of an open decision support tool that runs these scenarios, so planners can test a change before a storm rather than after.
Source: Peralta, J. C. A., Lim, M. T., and Reyes, R. C. (2023). Transport Network Efficiency during Typhoon Relief Operations. Philippine Journal of Science, 152(S1), 185 to 195