Key points are not available for this paper at this time.
This study investigates the dynamic degradation of Evapotranspiration (ET) time series induced by Xylella fastidiosa ( X.f. ) infection using the Complexity-Entropy Causality Plane (CECP). By evaluating MODIS ET datasets from infected sites (X2015, X2016, and X2017) against a healthy one (Matera), we characterize the “infection” signature as a fundamental transition from structured, deterministic dynamics toward stochastic disorder. Our findings reveal that infected signals undergo a simultaneous increase in permutation entropy ( H ) and a significant decrease in statistical complexity ( C ). This “loss of complexity” serves as a robust marker of ecosystem disturbance, overshadowing the underlying deterministic components of the ET signal. We demonstrate that with the embedding dimension d x = 4 there is a good diagnostic sensitivity, capturing intricate fourth-order temporal correlations with Mahalanobis distance D M > 1 . 3 and classification accuracy A U C = 0 . 83 . These results establish the H – C pair at d x = 4 as an optimal diagnostic tool for the detection of subtle anomalies in MODIS ET data, suggesting the CECP as a robust feature space for monitoring environmental signals, providing a reliable framework for the early detection and quantification of site-specific disturbances.
Telesca et al. (Sat,) studied this question.