Quantifying Population Exposure to Long-term PM10: A City-wide Agent-based Assessment

TL;DR

This study addresses the health effects of long-term PM10 exposure in Seoul using an in‑silico agent-based framework that extends Shin and Bithell’s approach to 25 districts and a 12-year horizon. It combines hourly PM10 data, origin–destination mobility, land-price–driven recovery, and age-structured demographics to quantify individual exposure and health decline when PM10 exceeds a threshold of $100 μg/m^3$, with health dynamics governed by a population-wide differential equation. Calibrating against Seoul hospital admissions from HIRA, the authors demonstrate that vulnerability emerges notably after about 7000 ticks (10 years), with greater impacts under a 3% annual pollution increase and for older and younger age groups, as well as in districts with road-proximate exposure. Scenario analyses reveal that resilience (adaptive capacity) and pollution trajectories jointly shape the at-risk population, highlighting spatial disparities and the need to test multiple thresholds and more detailed population-pollution dynamics for urban health planning.

Abstract

This study evaluates the health effects of long-term exposure to PM10 in Seoul. Building on the preliminary model Shin and Bithell (2019), an in-silico agent-based model (ABM) is used to simulate the travel patterns of individuals according to their origins and destinations. During the simulation, each person, with their inherent socio-economic attributes and allocated origin and destination location, is assumed to commute to and from the same places for 10 consecutive years. A nominal measure of their health is set to decrease whenever the concentration of PM10 exceeds the national standard. Sensitivity analysis on calibrated parameters reveals increased vulnerability among certain demographic groups, particularly those aged over 65 and under 15, with a significant health decline associated with road proximity. The study reveals a substantial health disparity after 7,000 simulation ticks (equivalent to 10 years), especially under scenarios of a 3% annual increase in pollution levels. Long-term exposure to PM10 has a significant impact on health vulnerabilities, despite initial resilience being minimal. The study emphasises the importance of future research that takes into account different pollution thresholds as well as more detailed models of population dynamics and pollution generation in order to better understand and mitigate the health effects of air pollution on diverse urban populations.

Figures (7)

• Figure 1: Official land prices in Seoul sub-districts in 2015 (a), and table of hypothetical health changes determined by land price in Korean currency ($1$\approx$₩1,200) (b)
• Figure 2: The implementation algorithm for each period. Retrieved and edited from Shin and Bithell shin2019
• Figure 3: Implementation in Netlogo 6.0.4 for Gangnam, Mapo, Gwanak, and Jongno
• Figure 4: Sensitivity of health loss and road proximity by time and risk population. The 3X2 array is indexed by districts and pollution weights to road patches. The decision threshold is indicated by the curly brackets. The decision threshold in both districts is between 0.003-0.005.
• Figure 5: The modelled result was calibrated against patient data. The observation data is based on a two-year average of the Korean CDC from 2015 to 2016