Detecting Air Pollution Clusters in Japan: A Spatial Analysis Approach

We rely on satellite data to study the spatial distribution of air pollutants and economic activity for 1,650 municipalities of all four main islands of Japan: Honshu, Kyushu, Hokkaido, and Shikoku. Specifically, we analyze atmospheric particulate matter and ozone concentrations, as well as population density, accessibility to cities, and night lights for the above islands. We then make use of principal component analysis, spatial dependence analysis, and methods of regionalization to endogenously classify the municipalities based on their similarity in attributes and geographic location. The spatial dependence analysis provides results which show us the specific sites where the high-value clusters (hot spots) and low-value clusters (cold spots) are located. These show a high positive correlation between economic activity and air pollution. Additionally, we perform a regionalization analysis of the variables under consideration, which specifies how the four main islands can be regionalized into six to nine geographical regions or structures, each. The regionalization takes into consideration both pollution levels and economic activity. We then conclude by discussing how these different analyses can complement each other, and how they contribute in finding the locations where policies related to air quality can help in improving the quality of life of the population.