Space-Time ARMA Models for Satellite Ozone Data

Abstract

In this paper, we develop some space-time autoregressive moving-average (ARMA) models for satellite ozone data, in which both the geographic neighbor effects and time-lag effects are considered. For the global TOMS ozone data over the 11-year period from January 1979 to December 1989, we divide the whole world into 10° latitude by 10° longitude blocks and calculate the monthly averages of total ozone for each geographic block. The space-time ARMA models then are applied to the monthly average series to assess the long-term changes in ozone concentrations. The long-term trend estimates over the world are all negative with the most negative trends occurring in the south polar latitudes, which is a reflection of the springtime Antarctic ozone hole that developed over the last several decades.

Computation for this paper were performed using computer facilities supported in part by National Sciences Foundation Grants number DMS 89-05292, DMS 87-03942 and DMS 86-01732 awarded to the Department of Statistics at the University of Chicago, and by the University of Chicago Block Fund. The first author was supported by National Aeronautics and Space Administration grant number NAG5-873. The second author was supported by National Science Foundation Grant DMS 89-02667