Breadcrumb
Periodically correlated spatial-temporal processes
Supervisor: Li Chen
Theme: Time Series
Periodicity is a natural phenomenon often observed from geophysical and environmental processes. For example, the hourly ozone concentrations observed at the US Environmental Protect Agency's Clean Air Status and Trends Network monitoring stations. At each station, the diurnal pattern may be observed in two ways. One is in the mean structure, such as the average for each of the 24 hours. The other is in the variation for each of the 24 hours. Moreover, the diurnal pattern may change from one location to another.
Research for periodically correlated random sequences has been done in univariate time series case, but there is little literature in the context of spatial-temporal statistics. Periodically correlated spatial-temporal processes are random processes in which there exists a periodic pattern that is much more complicated than periodicity in the mean structure. The periodicity in spatial-temporal covariance structure is particular of interest. However, the current spatial-temporal covariance models are incapable to present this important feature. The new periodic correlated spatial-temporal covariance models will be constructed upon the spectral representation. Both theoretical and practical aspects of periodically correlated spatial-temporal models will be investigated.