Searching for trends of change through exploratory data analysis of time series of remotely sensed images of SW Europe and NW Africa

Time series of remotely sensed imagery have proven to be a valuable tool to study the dynamics of the Earth surface. Some important features have been put in evidence in time series of greenness (measured as Normalized Difference Vegetation Index (NDVI)) fields at global and continental scales, and linked to changes in atmospheric variables and/or direct human impact. Analysing the same kind of information at regional scales is more challenging, as the increased detail implies increased complexity. We present a case of study of visualization coupled to statistical analysis based on Empirical Orthogonal Functions (EOF) of the seasonal pattern of greenness in SW Europe (52 degrees N, 11 degrees W to 30 degrees N, 5 degrees E) between 1999 and 2003. Our goal was to determine whether trends could be identified in the inter-annual change of some regions. We have found that for most of the area of study, inter-annual change in seasonal pattern of greenness consisted of an oscillation along a linear trajectory on the plane defined by the first two EO components. Nevertheless, a few scattered sites and two larger regions show trends, as both the direction and the sense of the inter-annual changes remained consistent from year to year during the period of study. Considering the dependency of the seasonal course of greenness on annual patterns of temperature and precipitation, the extent covered by these areas and the diversity of land cover that they include strongly suggest a response to changes in the seasonal patterns of key atmospheric variables for the same period. While the period of study is short, the particular response of these regions deserves further attention, including the analysis of longer series of NDVI and other satellite products, as well as the exploration of the actual relationships to atmospheric variables.