Global assessment of Vegetation Index and Phenology Lab (VIP) and Global Inventory Modeling and Mapping Studies (GIMMS) version 3 products
Earth observation-based long-term global vegetation index products are used by scientists from a wide range of disciplines concerned with global change. Inter-comparison studies are commonly performed to keep the user community informed on the consistency and accuracy of such records as they evolve. In this study, we compared two new records: (1) Global Inventory Modeling and Mapping Studies (GIMMS) normalized difference vegetation index version 3 (NDVI3g) and (2) Vegetation Index and Phenology Lab (VIP) version 3 NDVI (NDVI3v) and enhanced vegetation index 2 (EVI3v). We evaluated the two records via three experiments that addressed the primary use of such records in global change research: (1) leaf area index (LAI), (2) vegetation climatology, and (3) trend analysis of the magnitude and timing of vegetation productivity. Unlike previous global studies, a unique Landsat 30 m spatial resolution and in situ LAI database for major crop types on five continents was used to evaluate the performance of not only NDVI3g and NDVI3v but also EVI3v. The performance of NDVI3v and EVI3v was worse than NDVI3g using the in situ data, which was attributed to the fusion of GIMMS and MODIS data in the VIP record. EVI3v has the potential to contribute biophysical information beyond NDVI3g and NDVI3v to global change studies, but we caution its use due to the poor performance of EVI3v in this study. Overall, the records were most consistent at northern latitudes during the primary growing season and southern latitudes and the tropics throughout much of the year, while the records were less consistent at northern latitudes during green-up and senescence, and in the great deserts of the world throughout much of the year. These patterns led to general agreement (disagreement) between trends in the magnitude (timing) of NDVI over the study period. Bias in inter-calibration of the VIP record at northernmost latitudes was suspected to contribute most to these discrepancies.