A multitemporal and non-parametric approach for assessing the impacts of drought on vegetation greenness: a case study for Latin America

This study evaluates the relationship between the frequency and duration of meteorological droughts and the subsequent temporal changes on the quantity of actively photosynthesizing biomass (greenness) estimated from satellite imagery on rainfed croplands in Latin America. An innovative non-parametric and non-supervised approach, based on the Fisher-Jenks optimal classification algorithm, is used to identify multi-scale meteorological droughts on the basis of empirical cumulative distributions of 1, 3, 6, and 12-monthly precipitation totals. As input data for the classifier, we use the gridded GPCC Full Data Reanalysis precipitation time-series product, which ranges from January 1901 to December 2010 and is interpolated at the spatial resolution of 1° (decimal degree, DD). Vegetation greenness composites are derived from 10-daily SPOT-VEGETATION images at the spatial resolution of 1/112° DD for the period between 1998 and 2010.

The time-series analysis of vegetation greenness is performed during the growing season with a non-parametric method, namely the seasonal Relative Greenness (RG) of spatially accumulated fAPAR. The Global Land Cover map of 2000 and the GlobCover maps of 2005/2006 and 2009 are used as reference data to select study cases only on geographic areas that did not undergo land cover changes during the analysis period. The multi-scale information is integrated at the lowest spatial resolution available, i.e. 1° DD, and the impacts of meteorological drought episodes on seasonal greenness of rainfed crops are assessed at the regional scale. Final results suggest that the agricultural cycle at the regional scale is more correlated with long-standing and uninterrupted small timescale drought conditions that occur prior to vegetation growing season than with isolated and short long-term timescale drought events.