Outlier detection for geostatistical functional data: an application to sensor data

In this paper we propose an outlier detection method for geostatistical functional data. Our approach generalizes the proposal of Febrero et al. (2007, 2008) in the spatial framework. It is based on the concept of the kernelized functional modal depth that we have opportunely defined extending the functional modal depth. As an illustration, the methodology is applied to sensor data corresponding to long- term daily climatic time series from meteorological stations