Radiometric cross-calibration of Terra ASTER and MODIS

Calibration and validation play an essential role during the acquisition and processing of satellite data for Earth Observing System satellites in addition to being an integral part of maintaining scientific values of archived satellite data. The Advanced Spaceborne Thermal Emission and Reflection and Radiometer (ASTER) and Moderate Resolution Imaging Spectroradiometer (MODIS) are two of five sensors onboard the Terra platform. ASTER has a swath width of 60 km with 8 spectral bands in the visible and near infrared (VNIR) and thermal infrared (TIR) spectral range with a spatial resolution of 15-m (bands 1-3) and 90-m (bands 10-14), respectively while MODIS has a swath width of 2300 km with 36 spectral bands from visible to infrared spectral range with a spatial resolution of 250 m (bands 1-2), 500 m (bands 3-7), and 1 km (bands 8-36). ASTER is the ‘zoom’ lens and MODIS is the ‘keystone’ instrument for Terra; they provide quantitative measurements of various earth system variables to the scientific and to the broader community. The simultaneous view of the sensors simplifies the intercomparison between them and the current work relies on the use of the Railroad Valley Playa test site to reduce uncertainties caused by spatial heterogeneity and spectral differences in the sensors. The fact that Railroad Valley is a calibration test site for ASTER ensures that ASTER was tasked at a higher rate over this area providing more scenes for an intercomparison. The study compares ASTER L1B data for the three VNIR bands reprocessed with recent calibration updates and MODIS 02 Collection 6 data products for the similar bands. No correction for geometry angle is needed and coincident 3-km by 3-km regions are used to reduce the impact of spatial heterogeneity. A correction for spectral differences between the sensors is applied based on seasonal averages of EO-1 Hyperion spectral range. Results indicate that the calibrated radiance products from the two sensors agree to within the combined absolute uncertainties. There is no statistically significant temporal trend between the two and this should allow the correction between the two to within 0.5% over the 10 years studied here.