The impact of dataset selection on land degradation assessment

摘要: Abstract Accurate quantification of land degradation is a global need, particularly in the world’s dryland areas. However, there well-documented lack field data and long-term observational studies for most these regions. Remotely sensed offers only vegetation record that can be used assessment at national, continental or scale. Both rainfall datasets contain errors uncertainties, but little work has been done to understand how this may impact results. This study uses recently developed Time Series Segmented RESidual TREND (TSS-RESTREND) method applied six two assess dataset selection on estimates over Australia. Large differences methods produce precipitation did not significantly results with estimate average change varying by 95% On other hand, had much greater impact. Calibration Global Inventory Monitoring Modeling System Version 3 NDVI (GIMMSv3.0g) caused significant trends some Australia’s Though identified Australia, problematic calibration GIMMSv3.0g have effected values globally. These addressed updated GIMMSv3.1g which strongly recommended use future studies. Our analysis suggests using an ensemble composed multiple runs performed different allows identification cannot detected single run quality flags input datasets. A multi-run made provides more comprehensive uncertainty space time.