Comparisons of global land surface seasonality and phenology derived from AVHRR, MODIS, and VIIRS data

Land surface seasonality has been widely investigated from satellite observations for monitoring the dynamics of terrestrial ecosystems in response to climate change. A great deal of efforts has focused on the characterization of interannual variation and long‐term trends of vegetation phenological metrics derived from the advanced very high resolution radiometer (AVHRR) and the Moderate Resolution Imaging Spectroradiometer (MODIS) data across regional and global scales. Recently, Visible Infrared Imaging Radiometer Suite (VIIRS) data have become available for the detections of global land surface phenology. These data sets provide us a potential opportunity to generate a consistent long‐term climate data record of land surface phenology. However, to bridge the multiple‐sensor‐based phenology measurements, we need to understand their consistency and discrepancy across various geographical and ecological regions. To this end, we collected daily AVHRR, MODIS, and VIIRS data (~5 km) globally, compared their temporal trajectories of two band enhanced vegetation index (EVI2) during 2013 and 2014, examined their discrepancy of phenology retrievals, and explored the influence of EVI2 data quality from the three satellite sensors on phenology detections. The results revealed the similarity and discrepancy of EVI2 time series and land surface phenology retrievals globally. Specifically, EVI2 time series from VIIRS and MODIS observations were similar with some discrepancies mainly arising from unsystematic impact factors. In contrast, VIIRS EVI2 was systematically higher than AVHRR EVI2, in which their differences could be greatly reduced by intersensor calibration. Further, the quality of EVI2 time series among AVHRR, MODIS, and VIIRS varied largely across the globe, which was generally better in Northern Hemisphere than in Southern Hemisphere. Their differences in EVI2 data quality led to the inconsistencies in the detections of phenological dates. On average, the absolute difference of phenological transition dates among the three products was larger than 10 days in about 30% of pixels in Northern Hemisphere and more than 40% in Southern Hemisphere.