Robust detection of local outliers in multivariate spatial data

Multivariate spatial data are geographical locations on which non spatial variables are
measured. Such data may contain two types of outliers: global and/or local. Focus is here
on local outlier whose attribute values lie far from the values taken by its neighbors.
This poster has three main objectives. The first is to review some local detection
techniques that seem to perform well in practice. Secondly, an adaptation to one of these
is suggested to further develop its local characteristic. Then simulations based on Matérn
model are reported and discussed in order to compare in an objective way the different
detection techniques.