Inter-annual and seasonal trends of vegetation condition in the Upper Blue Nile (Abay) Basin: dual-scale time series analysis

摘要

A long-term decline in ecosystem functioning and productivity, often calledland degradation, is a serious environmental challenge to Ethiopia thatneeds to be understood so as to develop sustainable land use strategies.This study examines inter-annual and seasonal trends of vegetation cover inthe Upper Blue Nile (UBN) or Abbay Basin. The Advanced Very High ResolutionRadiometer (AVHRR)-based Global Inventory, Monitoring, and ModelingStudies (GIMMS) normalized difference vegetation index (NDVI) was used forlong-term vegetation trend analysis at low spatial resolution.Moderate Resolution Imaging Spectroradiometer (MODIS) NDVI data (MOD13Q1)were used for medium-scale vegetation trend analysis. Harmonic analyses andnon-parametric trend tests were applied to both GIMMS NDVI (1981–2006) andMODIS NDVI (2001–2011) data sets. Based on a robust trend estimator(Theil–Sen slope), most parts of the UBN (~ 77 %) showed apositive trend in monthly GIMMS NDVI, with a mean rate of 0.0015 NDVI units(3.77 % yr?1), out of which 41.15 % of the basin depictedsignificant increases (p < 0.05), with a mean rate of 0.0023 NDVIunits (5.59 % yr?1) during the period. However, the MODIS-basedvegetation trend analysis revealed that about 36 % of the UBN showed asignificant decreasing trend (p < 0.05) over the period 2001–2011 atan average rate of 0.0768 NDVI yr?1. This indicates that the greeningtrend of the vegetation condition was followed by decreasing trend since themid-2000s in the basin, which requires the attention of land users anddecision makers. Seasonal trend analysis was found to be very useful toidentify changes in vegetation condition that could be masked if onlyinter-annual vegetation trend analysis was performed. Over half (60 %) ofthe Abay Basin was found to exhibit significant trends in seasonality overthe 25-year period (1982–2006). About 17 and 16 % of the significanttrends consisted of areas experiencing a uniform increase in NDVI throughoutthe year and extended growing season, respectively. These areas were foundprimarily in shrubland and woodland regions. The study demonstrated thatintegrated analysis of inter-annual and intra-annual trends based on GIMMSand MODIS enables a more robust identification of changes in vegetation condition.