Amending Method for Multi-Dimensional Time Series with Heavy Missing Data and Its Application in Environment Monitoring

ObjectivesTo explore a better amending method for a set of correlated time series with heavy missing data and its application to monitoring data on environment pollution.MethodsAmodel of Simultaneous Linear Equations with ARMA Error is proposed for iterative estimation of missing values.Simulation study is performed on the basis of an AR(1) model with input series and applied study is performed on thbasis of a real montoring data set on CO-pollution respectively.The results are compared with those estimated wih three usually used apporaches,Multivariate Linear Regression,Simple ARIMA and interpolation by cubic spline in terms of statistical property.Programming is performed with statistical software SAS.ResultsThe iterative approach based on the model of Simultaneous Linear Equations with ARMA Error is better than other three approaches significantly.ConclusionThe amending method proposed can be used to estimate the missed values of a set of multi-dimensional time series with heavy missing,and apply to relevant monitoring data in real life.