Global Analysis of the Differences Between the MODIS Vegetation Index Compositing Date and the Actual Acquisition Date


Image compositing enables the generation of continuous and equidistant time series of vegetation indices (VIs) as a requisite for deriving phenological metrics. This is achieved by selecting a representative choice from available observations within a compositing period. Commonly, phenological processors associate the VI value with a reference date rather than the actual acquisition date. This letter focuses on a global assessment of the resulting temporal discrepancy between the assumed reference date and the actual acquisition date using Moderate Resolution Imaging Spectroradiometer (MODIS)-VI day of composite (DOC) information from 2000 to 2015. It is shown that the long-term global mean DOC coincides with the midday of a compositing period (day 8.44 ± 0.44). The global summarized 2-D frequency distributions of succeeding compositing periods revealed that only 16.93% of all occurrences indicate a 16-day difference. The commonly used assumptions that the DOC coincides with the first, mid, or last day of the compositing period on average only hold true for 7.02%, 6.07%, and 6.03% of temporal succeeding compositing periods, respectively. Spatial patterns of 2-D frequency distributions occur as a result of climatic regions and seasonality. A MODIS tile is on average composed of more than 50% DOC combinations, indicating deviations (1-31 days) from the 16-day difference. Since information about the actual acquisition date is available, and methods for the incorporation of irregular data into phenological analysis do exist, it is recommended to incorporate this information into the phenology research based on VI time series to avoid unnecessary uncertainty.

IEEE Geoscience and Remote Sensing Letters