Abstract
Accurate and timely land surface phenology (LSP) provides essential information for investigating the responses of terrestrial ecosystems to climate changes and quantifying carbon and surface energy cycles on the Earth. LSP has been widely investigated using daily Visible Infrared Imaging Radiometer Suite (VIIRS) or Moderate Resolution Imaging Spectroradiometer (MODIS) observations, but the resultant phenometrics are frequently influenced by surface heterogeneity and persistent cloud contamination in the time series observations. Recently, LSP has been derived from Landsat-8 and Sentinel-2 time series providing detailed spatial pattern, but the results are of high uncertainties because of poor temporal resolution. With the availability of data from Advanced Baseline Imager (ABI) onboard a new generation of geostationary satellites that observe the earth every 10–15 min, daily cloud-free time series could be obtained with high opportunities. Therefore, this study investigates the generation of synthetic high spatiotemporal resolution time series by fusing the harmonized Landsat-8 and Sentinel-2 (HLS) time series with the temporal shape of ABI data for monitoring field-scale (30 m) LSP. The algorithm is verified by detecting the timings of greenup and senescence onsets around north Wisconsin/Michigan states, United States, where cloud cover is frequent during spring rainy season. The LSP detections from HLS-ABI are compared with those from HLS or ABI alone and are further evaluated using PhenoCam observations. The result indicates that (1) ABI could provide ~3 times more high-quality observations than HLS around spring greenup onset; (2) the greenup and senescence onsets derived from ABI and HLS-ABI are spatially consistent and statistically comparable with a median difference less than 1 and 10-days, respectively; (3) greenup and senescence onsets derived from HLS data show sharp boundaries around the orbit-overlapped areas and shifts of ~13 days delay and ~15 days ahead, respectively, relative to HLS-ABI detections; and (4) HLS-ABI greenup and senescence onsets align closely to PhenoCam observations with an absolute average difference of less than 2 days and 5 days, respectively, which are much better than phenology detections from ABI or HLS alone. The result suggests that the proposed approach could be implemented the monitor of 30 m LSP over regions with persistent cloud cover.
Highlights
Land surface phenology (LSP) plays an important role in understanding the response of terrestrial ecosystems to environmental changes [1,2]
The number of high-quality observations for northern forest area was larger than central wetland area either for Advanced Baseline Imager (ABI) or harmonized Landsat-8 and Sentinel-2 (HLS) data (Figure 2a,b)
Differences in HLS observations were apparent around the satellite overpass-orbits where pixels within the area could have up to 13 observations while the neighboring pixels only had less than 5 or even overpass-orbits where pixels within the area could have up 3 observations (Figure 2b)
Summary
Land surface phenology (LSP) plays an important role in understanding the response of terrestrial ecosystems to environmental changes [1,2]. Fine spatial resolution satellite data, such as Landsat-8 as well as Sentinel-2A, Sentinel-2B data, are capable of characterizing vegetation seasonality and phenology properties at a field scale (30m) [20,21,22,23,24]. At such spatial extent, the phenological behaviors of individual land cover type (i.e., specific crop type or tree species) could be revealed [25]. Despite efforts have been put on the generation of harmonized
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
