Improvement of land surface phenology monitoring by fusing VIIRS observations with GOES-16/17 ABI time series

Abstract

Land Surface Phenology (LSP) has been widely derived from polar-orbiting satellite observations to characterize terrestrial vegetation dynamics. However, the uncertainty of LSP detections over large areas is always a big concern because of cloud contamination in the satellite time series, particularly in persistently cloudy regions. The Advanced Baseline Imager (ABI) onboard Geostationary Operational Environmental Satellite-R (GOES-R) provides a high likelihood of obtaining cloud-free observations throughout the vegetation growing season due to the high temporal resolution of 10 minutes. Therefore, this study investigated LSP detections at 500 m pixels from VIIRS (Visible Infrared Imaging Radiometer), ABI, and fused VIIRS-ABI time series in 2019 over North America between 12°N and 48°N. Specifically, the 3-day composite VIIRS NBAR (Nadir Bidirectional Reflectance Distribution Function (BRDF)-Adjusted Reflectance) EVI2 (two-band Enhanced Vegetation Index) time series was first generated. Similarly, the 3-day composite ABI EVI2 time series was also produced after performing BRDF-adjustment of 10-min GOES-16/17 ABI surface reflectance. The 3-day VIIRS EVI2 time series was then fused with ABI EVI2 observations to generate the synthetic high spatiotemporal VIIRS-ABI EVI2 time series. Further, LSP was separately detected from the VIIRS, ABI, and fused VIIRS-ABI EVI2 time series. Finally, the three LSP detections were analyzed with the variation of cloud cover and ABI view zenith angle (VZA) and validated using the LSP reference produced from the fusion of Harmonized Landsat 8 and Sentinel-2 (HLS) observations with PhenoCam time series. The results showed that VIIRS-ABI LSP could overcome the limitations in the LSP detections from either VIIRS or ABI alone. The improvement of VIIRS-ABI LSP could be over 15 days relative to ABI LSP in large VZA regions and 5 days relative to VIIRS LSP in regions prone to persist cloud cover. Because of the high complementarity between the polar-orbiting and geostationary satellites, their fusion could significantly improve the generation of global LSP products.