Author:

Keywords:

decomposition, adaptive analysis, time series analysis, non-stationarity, remote sensing, Social Sciences, Science & Technology, Physical Sciences, Economics, Mathematics, Interdisciplinary Applications, Social Sciences, Mathematical Methods, Statistics & Probability, Business & Economics, Mathematics, Mathematical Methods In Social Sciences, HILBERT SPECTRUM, VARIABILITY, PRODUCTS

Abstract:

Vegetation monitoring by satellite sensors has delivered 30-year time series of vegetation cover images over large areas. Decomposition of a pixel’s Vegeta-tion Index (VI) time series reveals the underlying processes of vegetation cover change at various time scales. Ensemble Empirical Mode Decomposition (EEMD) is a data-adaptive technique to isolate the effects of non-stationary recurrent climatic variability. To recognize significant patterns in the detected components, we propose a local significance test. We tested the method’s accuracy and sensitivity on a set of synthetic time series that represent our knowledge of climatic phenomena and vegeta-tion dynamics. It was also demonstrated for a set of real VI time series over a study area in East and Central Africa.