Evaluation of sensor calibration uncertainties on vegetation indices for MODIS

The impact of reflectance calibration uncertainties on the accuracies of several vegetation indices (VI's) was evaluated for the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor onboard the TERRA platform. A set of uncertainty propagation equations were designed to model the propagation of calibration uncertainties from top-of-atmosphere (TOA) reflectances to atmospherically-corrected VI's, The soil-adjusted vegetation index (SAVI), the atmospherically-resistant vegetation index (ARVI), and the enhanced vegetation index (EVI) were evaluated along with the normalized difference vegetation index (NDVI). The resultant VI uncertainties associated with calibration mu(cal) varied with both surface reflectances and atmospheric conditions. Uncertainties in the NDVI and ARVI were highly dependent on pixel brightness, with the largest uncertainties occurring over dark targets with little or no vegetation, The SAVI uncertainties were nearly constant throughout a range of target brightness and vegetation abundance, The EVI uncertainties linearly increased with increasing EVI values. Atmosphere turbidities increased calibration uncertainties in all the VI's through their effect on TOA reflectances. The VI uncertainties were also found to decrease when the calibration errors were positively correlated between bands, Using field observational canopy reflectance data, the mean VI uncertainties were estimated to be +/-0.01 VI units for the NDVI and SAVI, and +/-0.02 VI units for the ARVI and EVI under normal atmosphere conditions (greater than or equal to 20 km visibility) and for a 2% reflectance calibration uncertainty. The higher uncertainties in the ARVI and EVI resulted from the inclusion of the blue band, adding to the overall calibration uncertainty budget, Overall, the magnitudes of the mean VI uncertainties due to sensor calibration were less than 2% of VI dynamic range and can be considered satisfactorily low.