Spring phenology and phenology-climate links inferred from two remotely sensed vegetation indices across regions and biomes

Abstract. The timing of spring greenup (SG) as inferred by remotely sensed vegetation indices have showed contrasting dynamics across the same region and periods. Assessing the uncertainty in SG associated with different Normalized Difference Vegetation Index (NDVI) products is essential for robustly interpreting the links between climate and phenological dynamics. We compare SG inferred from two NDVI products over the period 2001–2013: (1) Terra Moderate Resolution Imaging Spectroradiometer (MODIS) and (2) National Oceanic and Atmospheric Administration's (NOAA's) Advanced Very High Resolution Radiometer (AVHRR) instruments processed by the Global Inventory Monitoring and Modeling Studies (GIMMS) to explore confidence and uncertainty in the NDVI-inferred SG trend and its links to climate variability. Both MODIS and GIMMS agreed in showing an advancement of SG in northern Canada, the eastern United States, and Russia, as well as a delay in SG in western North America, parts of Baltic Europe and East Asia. In the regions with advanced SG, GIMMS inferred much weaker advancement whereas in the regions with delayed SG, GIMMS inferred much stronger delay than MODIS. This resulted in a GIMMS SG delay in both North America and Eurasia. MODIS data show no significant SG shift in North American for spatial heterogeneity in SG shift, but dominant SG advancement in Eurasia. The SG advancement inferred from MODIS is associated with a stronger coupling between SG and temperature and a stronger sensitivity across biomes as compared to GIMMS. The main uncertainty in the SG trend and SG-temperature sensitivity are in northern high latitudes (>50° N) where GIMMS and MODIS show different magnitude and sign of the annual SG anomalies. Compared to 1988–2000, inter-biome GIMMS SG-temperature sensitivity is stable and the SG-temperature sensitivity increased in the boreal and Arctic biomes despite a slight reduction in the SG-temperature coupling over the period 2001–2013. The explanation for the increased SG-temperature sensitivity remains unclear and requires further investigation. We suggest broader evaluation of the NDVI products against field measurements and inter-validation for robust assessment of vegetation dynamics.