Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Hydrology and Remote Sensing Laboratory » Research » Publications at this Location » Publication #329120
Title: Exploration of scaling effects on coarse resolution land surface phenology

Author
item ZHANG, XIAOYANG - South Dakota State University
item WANG, JIAMIN - South Dakota State University
item Gao, Feng
item LIU, YAN - University Of Massachusetts
item SCHAAF, CRYSTAL - University Of Massachusetts
item HENEBRY, G. - South Dakota State University
item FRIEDL, MARK - Boston University
item YU, YUNYUE - National Oceanic & Atmospheric Administration (NOAA)
item JAYAVELU, S. - South Dakota State University
item GRAY, JOSHUA - Boston University
item LIU, LINGLING - South Dakota State University
item YAN, DONG - South Dakota State University

Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/7/2017
Publication Date: 1/13/2017
Citation: Zhang, X., Wang, J., Gao, F.N., Liu, Y., Schaaf, C., Henebry, G., Friedl, M., Yu, Y., Jayavelu, S., Gray, J., Liu, L., Yan, D. 2017. Exploration of scaling effects on coarse resolution land surface phenology. Remote Sensing of Environment. 190:318-330.

Interpretive Summary: Land surface phenology (LSP) affects the rate of photosynthesis and plant water use. It can be detected from a dense time-series of remote sensing data. There are a number of LSP data products available today at coarse spatial resolution (a few hundred meters or coarser). However, biophysical characteristics of LSP detected from coarse resolution remote sensing data have not been well understood. This paper examines LSP biophysical characteristics by comparing detections from coarse resolution to fine resolution. Results show that the start of the growing season from coarse resolution (375m) were comparable to Landsat resolution (30m) in central Iowa. The overall differences were less than 5 days. Understanding LSP detected from different spatial resolutions is important for monitoring crop conditions required by the National Agricultural Statistics Service and Foreign Agricultural Service for more accurate yield assessments and predictions.

Technical Abstract: A great number of land surface phenoloy (LSP) data have been produced from various coarse resolution satellite datasets and detection algorithms across regional and global scales. Unlike field- measured phenological events which are quantitatively defined with clear biophysical meaning, current LSP detections only determine the timing of variations in satellite greenness. Since activities to bridge LSP and field observations are very challenging and limited, our understanding of LSP biophysical characteristics is very poor. Therefore, we explored LSP biophysical characteristics in the start of growth season (SOS) by comparing detections from coarse resolution datasets with those from fine resolution imagery. Specifically, using hybrid piecewise-logistic-model-based LSP detection algorithm (HPLM-LSPD), we first detected SOS at a field scale (30m) from the reflectances fused by MODIS data and Landsat 8 OLI observations (termed as OLI SOS here) and at a resolution of 500m from the Visible Infrared Imaging Radiometer Suite (VIIRS) observations, respectively, in central Iowa, the United States. The results revealed the SOS complexities and biophysical characteristics at coarse resolutions. Specifically, OLI SOS variation defined using standard deviation (SD) was as large as 40 days within a very heterogeneous VIIRS pixel while it was less than 10 days in more than 40% of the VIIRS pixels. Further, VIIRS SOS was well represented by more than 60% of OIL SOS values within a homogeneous VIIRS pixel but only by less than 20% of OIL pixels within a very heterogeneous VIIRS pixel. Moreover, SOS in a coarse resolution pixel reflected the timing at which 30% of area had turned green across various heterogeneous levels. This result also suggests that SOS at a coarse resolution should be well aggregated by selecting the timing at 30% percentile SOS at fine resolution. Finally, it was demonstrated that VIIRS SOS was comparable with OLI SOS with an overall difference of less than 5 days in homogeneous regions.