초록 close

A Markov model is presented to explain the behavior of PM10 concentrations in Seoul, exploring its predictive capabilities. This article focuses on (1) constructing a transition probability matrix, representing the movements of daily PM10 concentrations during three months from March to May, 2001~2005, (2) evaluating the validity of the transition probability matrix and testing of its stationarity, and (3) analyzing its practical application to the existing PM10 warning system. The statistical tests show that the three-month transition matrix for each year includes a stationary pattern, which verify its predictive capabilities for the time period. Then, the further analyses show its steady-state probabilities and the expected value of first passage time for each PM10 level. The results imply that the transition probability matrix and subsequent Markov process analyses can be very useful to forecast the pollution level in a simple way.