Terra MODIS Band 27 Electronic Crosstalk Effect and Its Removal

Abstract

The MODerate-resolution Imaging Spectroradiometer (MODIS) is one of the primary instruments in the NASA Earth Observing System (EOS). The first MODIS instrument was launched in December, 1999 on-board the Terra spacecraft. MODIS has 36 bands, covering a wavelength range from 0.4 µm to 14.4 µm. MODIS band 27 (6.72 µm) is a water vapor band, which is designed to be insensitive to Earth surface features. In recent Earth View (EV) images of Terra band 27, surface feature contamination is clearly seen and striping has become very pronounced. In this paper, it is shown that band 27 is impacted by electronic crosstalk from bands 28-30. An algorithm using a linear approximation is developed to correct the crosstalk effect. The crosstalk coefficients are derived from Terra MODIS lunar observations. They show that the crosstalk is strongly detector dependent and the crosstalk pattern has changed dramatically since launch. The crosstalk contributions are positive to the instrument response of band 27 early in the mission but became negative and much larger in magnitude at later stages of the mission for most detectors of the band. The algorithm is applied to both Black Body (BB) calibration and MODIS L1B products. With the crosstalk effect removed, the calibration coefficients of Terra MODIS band 27 derived from the BB show that the detector differences become smaller. With the algorithm applied to MODIS L1B products, the Earth surface features are significantly removed and the striping is substantially reduced in the images of the band. The approach developed in this report for removal of the electronic crosstalk effect can be applied to other MODIS bands if similar crosstalk behaviors occur.

Long‐term drift induced by the electronic crosstalk in Terra MODIS Band 29

“…The crosstalk effect is strongly detector dependent as demonstrated by the crosstalk coefficients and the detector difference before crosstalk effect correction. After the crosstalk correction is applied, the detector differences become much smaller and the This has also been demonstrated in Terra bands 27 and 28 [Sun et al, 2014a[Sun et al, , 2014b[Sun et al, , 2015 as mentioned previously. In other words, the higher-order effect should be relatively small and, currently, no clear evidence shows that higherorder effects have significant contributions to the crosstalk effect.…”

“…However, band 31 does not saturate for an entire lunar image. With an assumption that there is a linear relationship between DNs of any of bands 27-30 and those of band 31, the DNs of the band in the saturated area can be estimated from the DNs of band 31 and the ratios of the two bands' DNs in the unsaturated region [Sun et al, 2014a]. From Figure 1, a tiny protrusive feature is observed near the bottom of the oval cylindrical structure.…”

“…For a MODIS TEB, the crosstalk contribution from other bands can be approximated by the expression [Sun et al, , 2014a…”

“…The crosstalk may induce striping due to the detector dependence of the crosstalk effect [Sun et al, 2014a[Sun et al, , 2015. For Terra band 29, the crosstalk effect indeed varies with the detector as demonstrated by the crosstalk coefficients in Figures 3-5.…”

“…K effect in Earth View (EV) Brightness Temperature (BT) for bands 27 and 28, respectively, over the past 15 years[Sun et al, , 2014a[Sun et al, , 2014b[Sun et al, , 2014cMadhavan et al, 2014]. Recent vicarious ground and inter sensor comparisons with ASTER have shown that Terra MODIS band 29 has a temporal bias in measurements around the typical radiance levels[Hook et al, 2014].…”