Detection of environmental catastrophes

Abstract

For on-line detection of environmental catastrophes spatial as well as temporal effects are crucial. Observations of some substance are made at different locations in space at discrete times. Mainly two spatial features are considered: first, a situation where the observations made at the same time but at different sites are correlated, and secondly, a situation where the catastrophe originates from a source with known position in space and then spreads, passing other known positions as time passes. The first situation turns out to be a special case of a previously studied situation in multivariate surveillance. The second case is reduced to a univariate problem and a brief evaluation of the Shewhart method, the Cusum method and the Shiryaev-Roberts method is made. The theory is applied on the case of surveillance of radiation. The suggested methods are compared with the method presently in use in Sweden.