Advances in satellite Earth observation have resulted in the development of consistent
global historical records of environmental and climatic variables, forming enormous amounts
of multivariate time series. In this work we present a novel machine learning framework for
detecting relationships between climatic time series and vegetation indices. Our pipeline
consists of several components, including data fusion from various databases, time series
decomposition techniques, feature construction methods and predictive modeling …

For radiation two different products have been collected, the first one based on satellite data
(NASA/GEWEX Surface Radiation Budget (SRB)[21]) and the second one on reanalysis
data (ERA-Interim)[3]. The surface soil moisture products have been produced by satellite
data (Global Land Evaporation-Amsterdam Methodology (GLEAM)[14], NASA [15], Climate
Change Initiative (CCI)[11, 12, 24]). The three soil moisture products by CCI consist of a
merged product created from all active data sets, a merged product created from all passive …