Pesticide concentration monitoring: Investigating spatio‐temporal patterns in left censored data

Monitoring pesticide concentration is very important for public authorities given the major concerns environmental safety and likelihood increased health risks. An aspect of this process consists in locating abnormal signals, from a large amount collected data. This kind data usually complex since it suffers limits quantification leading to left censored observations, sampling procedure which irregular time space across measuring stations. The present manuscript tackles precisely issue detecting spatio-temporal collective anomalies levels, introduces novel methodology dealing with heterogeneity. latter combines change-point detection applied series maximum daily values all stations, clustering step aimed at spatial segmentation Limits are handled procedure, by supposing an underlying left-censored parametric model, piece-wise stationary. Spatial takes into account geographical conditions, may be based on river network, wind directions, etc. Conditionally temporal segment cluster, one eventually analyse identify contextual anomalies. proposed illustrated detail set containing prosulfocarb levels surface waters Centre-Val de Loire region.