Evaluation of six satellite-derived Fraction of Absorbed Photosynthetic Active Radiation (FAPAR) products across the Australian continent

Satellite remote sensing products of the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) are routinely used for diverse applications in Earth-System and land-surface modelling and monitoring. The availability of numerous products creates a need to Understand the level of consistency between products, and reasons for inconsistencies. We evaluate the consistency of six FAPAR products (MODIS, MERIS, SeaWIFS, MODIS-TIP, SPOT-VEG, and AVHRR) across the Australian continent, using multi-year records. We find that seemingly large differences in FAPAR products over much of Australia can be explained by a simple offset present in certain products. Additional inconsistencies arise from different sensitivities in FAPAR to changes in vegetation cover. These inconsistencies can in turn be partially attributed to changes in biome type that are relevant to certain products and related model-specific assumptions. The satellite FAPAR products are compared to similar to 800 observation-based estimates of fractional vegetation cover at field sites across Australia. After accounting for offsets in FAPAR, relatively high agreement occurs at sites classified as grasslands, shrublands and managed land (agriculture). Significant disagreement occurs at sites correctly classified as forests. Consequently, some products show significant differences in FAPAR between regions of similar vegetation cover but different biome classification. We find that all products show a much lower sensitivity to fractional vegetation cover (range in coefficient of linear regression: 0.28-0.61) than is predicted theoretically (0.96-1.18) using a canopy radiative transfer model. Reasons for this discrepancy are discussed. (C) 2013 Elsevier Inc. All rights reserved.