Review of Terra MODIS thermal emissive band L1B radiometric performance

The MODerate-resolution Imaging Spectroradiometer (MODIS) on NASA’s Earth Observing System Terra satellite, launched into orbit on 18 December 1999, will have a “first light” 15th anniversary on 24 February 2015. For nearly15 years the MODIS instrument has provided radiances in all spectral bands. Though some detectors have fallen below SNR thresholds, the vast majority of spectral bands continue to provide high quality L1B measurements for use in L2 science algorithms supporting global climate research. Radiometric accuracy of the Terra MODIS thermal emissive bands (TEBs) in the C6 L1B product has been assessed using various approaches over the nearly 15 year Terra MODIS data record, including comparisons with instruments on the ground, in aircraft under-flights, and on other satellites. All of these approaches contribute to the understanding of the Terra MODIS radiometric L1B performance. Early in the lifetime of Terra, ground-based measurements and NASA ER-2 aircraft under-flights revealed that TEBs in the infrared window (“window” bands) are well calibrated and performing within accuracy specifications. The ER-2 under-flights also suggested that many atmospheric bands may be performing outside of specification, especially LWIR CO2 sensitive bands that are subject to optical crosstalk, although analysis uncertainties are larger for atmospheric bands. Beginning in 2007, MetOp-A IASI observations were used to evaluate Terra MODIS TEB performance through Simultaneous Nadir Overpass (SNO) comparisons. These inter-satellite comparisons largely affirm the early aircraft and ground-based evaluations, showing that all Terra MODIS window bands have small biases, minimal trending, and minor detector and mirror side striping over the 2007-2013 timeframe. Most atmospheric bands are performing satisfactorily near to specification; however, biases, striping and trending are large and significantly out of specification in the water vapor sensitive band 27 and ozone sensitive band 30 while the CO2 sensitive band 36 bias significantly exceeds specification. The investigation has found that an effective spectral shift significantly reduces biases and scene temperature dependence (but not trends) in most atmospheric bands, bringing them closer to, if not within, specification.