TIME SERIES MODEL APPLIED TO ENVIRONMENTAL MONITORING DATA ANALYSES

摘要

Time series data often arise when monitoring the degree of environmental pollution in a target zone. The purpose of usage of time series properties is to understand the driving forces and structures that produce the observed data and to fit the data into a model and to proceed them to forecasting, monitoring or even feed back and feed forward control. The essential difference between modelling data via time series methods and using the process monitoring methods is that in time series analysis the data points taken over time may have internal structure such as autocorrelation, trend or seasonal variation. This paper uses some of the time series techniques for modelling and analysis of the CO monitoring data measured by the Agency for Environmental Protection, Bucharest in some important cross-roads of the city. The hourly registration of CO concentrations in the air forms a univariate time series that consists in single scalar observations recorded sequentially over equal time increments. This time series display periodicity that represents sinusoidal fluctuations over 24 h and also over entire week and seasonality of the year. The seasonal pattern was consistent with a wave sine model. The rescaled adjusted range analysis is used for characterising the dependence of the properties on the time series.