Analyses of pre-launch test data showed that radiometric errors equal to 10% or more had been present in the SWIR bands (Bands 5, 6, 7 and 26), and 5% or more in the MWIR bands (Bands 20 - 25).
All 500 m bands are sampled twice (in scan direction) for each instance that the 1000 m bands are sampled, and these are identified as sub-frame 1 and sub-frame 2. For SWIR Bands 5, 6 and 7, there are differences in radiometric performance between the two sub-frames that we believe to be primarily due to the presence of electronic cross-talk.
Impact to L1B Data Set
This effect exists for all L1B products produced using software Version 3.0 and lower and amounts to radiometric errors of up to 10% in the SWIR bands (5, 6, 7, 26) and 5% in the MWIR bands (20 – 25).
Estimated Impact to L2 Science Products
This effect exists for all L1B products produced using software Version 3.0 and lower and is associated with errors greater than 2% in the SWIR bands and greater than 1% (0.75% fo band 20) for the MWIR bands.
Cause of Effect
These errors are due to electronic cross-talk while viewing structured Earth scenes, and were not evident in the low contrast scenes used for calibration. Electronic cross-talk is believed to be related to incomplete restoration of the baseline voltage on the capacitor used for signal integration in the FPA electronics. This effect was reduced by means of a change in the electronics external to the FPA prior to launch, but the system was not re-tested between this electronics change and launch.
Mitigation Approach/ Added Testing to Characterize Effect
Early MODIS data indicated that the electronics change reduced the errors from electronic cross-talk, but did not eliminate the effects. Subsequent on-orbit testing indicated that changing the detector bias voltage would reduce this cross-talk effect. For bias voltages larger than the –1.1 V used at launch, these changes led some detectors into reverse bias operation. Beginning in November 2000, a bias voltage level was found that minimized the electronic cross-talk and allowed all detectors to operate in reverse bias. Quantitative results for the magnitude of this effect are not available at the present time.
The reflective LUTs V184.108.40.206 contain a B-side SWIR OOB (5.3 micron leak) correction to improve the calibration of the firsrt subsample (in bands 5, 6, 7) and for band 26. The reflective LUTs V220.127.116.11 incorporate a new SWIR OOB correction. In both cases, the correction switch is On for the B-side calibration, OFF for the A-side. The SWIR OOB correction switch had previously been turned OFF for SWIR bands in LUT V18.104.22.168.
Estimated Schedule to Complete
Using SRCA tests, the magnitude of this effect was evaluated for the period after November 1, 2000. The analysis review is scheduled to be completed by October 31, 2001.
History of Characteristic
Time-dependent table of Electronic cross-talk testing