A fuzzy anomaly indicator for environmental monitoring at continental scale

Environmental status assessment and monitoring can be performed by the integration of multi-source datasets at continental and global scales. We propose a methodology for the development of a new anomaly indicator (AI) which can highlight the occurrence of anomalous conditions in a synthetic fashion by analysis of a set of spatial input data. Anomalous conditions are defined relative to long-term average assumed as normal or reference status of the vegetated land surface. The indicator is defined according to fuzzy set theory which is a powerful means of handling uncertain and imprecise knowledge of environmental systems. The indicator integrates, in an innovative way, the anomaly scores of a set of contributing factors extracted from the analysis of historical time series, mainly of Earth observations data. These time series are used to automatically derive the fuzzy membership functions that quantify the contribution of each factor to the final indicator. No reference data and expert knowledge are strictly required for the implementation of the AI although the methodology allows customization where this type of information is available. The method was tested over the African continent for the period 1996-2002; monthly AI values were derived with input datasets of vegetation phenology and rainfall estimates. The output AI continental maps bring new information by integrating multiple factors and they highlight patterns of anomalous conditions of the status of the environment. The analysis of the correlation with the El Niño Southern Oscillation (ENSO) shows that the AI is able to identify the effects of this phenomenon and its spatio-temporal dynamics. The 1997-1998 and 2000-2001 ENSO events are clearly highlighted by the highest AI values in specific regions of the continent. The indicator proposed is a valuable tool which can help guide in depth and detailed investigations of environmental conditions at local scale.