Assessment of health risk due to PM10 using fuzzy linear membership kriging with particle swarm optimization

Abstract

Air quality is an important determinant of individual as well as broader well-being. Major pollutants include gasses as well as assorted suspended particulate matter (PM). In this paper, we focus on PM10, which are a collection of particles with median aerodynamic diameter less than 10 m that remains suspended in the air for long periods. PM10, usually consist of smoke, dirt and dust particles, as well as spores and pollen, could easily be inhaled deep into lung. As a result, high outdoor PM10 concentration poses significant health hazard, and accurate modeling and prediction of health risk due to PM10 assume importance in pollution and public health management. In this backdrop, we propose an improved health risk assessment technique, and demonstrate its efficacy using widely used California PM10 database. At the heart of the proposed method lies indicator kriging, a well-known risk estimation technique. However, improved assessment of subjective health risk is achieved by posing the problem in a fuzzy setting, and optimizing the associated membership functions. In particular, we employ particle swarm optimization (PSO) algorithm, which has been motivated by natural behavior of organisms such as fish-schooling and bird flocking, and proven effective in various optimization contexts. We apply the fuzzy PSO membership grade Kriging technique to predict the PM10 spatial distribution over the entire California state.