Steinberg, DC, Goetz, S (2009). Assessment and extension of the MODIS FPAR products in temperate forests of the eastern United States. INTERNATIONAL JOURNAL OF REMOTE SENSING, 30(1), 169-187.
Spatial and temporal variability in the quality of the collection 4 and 5 MODIS FPAR products were examined over a 1.1 million km2 region dominated by temperate forests. The MODIS FPAR products were parsed into different quality levels based on the retrieval method of the FPAR estimates, and the spatial coverage of the different quality MODIS subsets was examined through time and by land cover. The spatial extent of good quality FPAR estimates, derived from the main radiative transfer algorithm without saturation, was found to be quite limited, particularly during the growing season. The majority of the MODIS C4 FPAR retrievals were derived from the back-up, empirically based algorithm. The C5 products available to date showed greater data quality but still had limitations in forested areas. As an alternative to using the lower-quality MODIS FPAR data, we explored the utility of estimating FPAR from the more spatially extensive (and higher resolution) MODIS NDVI product using a simple, regionally based linear regression. A strong linear relationship between high quality MODIS NDVI and FPAR products within our study site was found, indicating its utility for estimating FPAR at regional scales where good quality retrievals from the MODIS algorithm are not available and where higher resolution FPAR maps are desired. We discuss these results and their significance for down-stream MODIS users.