Time Series Modeling of Urban Pollution Levels.
Final rept. Oct 73-Nov 74,
SPACE AND MISSILE SYSTEMS ORGANIZATION NORTON AFB CALIF
Pagination or Media Count:
Research was conducted to find enough time series data of various types of signals to display the versatility of the modeling technique called Autoregressive-Moving Average ARMA p,q. This was done by obtaining several 24-hour average air pollutant measurements from Bayonne, New Jersey, and several hourly water quality measurements from the Great Miami River, Dayton, Ohio. This data was chosen because of the large amount available and the ease with which it could be handled on a computer. Given these sets of n observations from the time series assumed here to be stationary, the procedure for finding an adequate ARMA p,q model to represent the series can be described as an iterative process involving three fundamental steps 1 identification of model to be used, 2 estimation of model parameters, and 3 checking the candidate model by an analysis of the residuals. These models offer a compact descriptive format in which to store vast amounts of observed data. In addition, the fitted models give insight into the nature of the signals, the problems they cause, and possible solutions.
- Statistics and Probability
- Air Pollution and Control