This effect appears in the MWIR thermal bands (Bands 20 - 25). It is seen as along track striping within a scene. [VERIFY: It appears less prominently in the LWIR PV bands (27 – 30)]
Illustration of effect
Band 21 MODIS scene shows striping without crosstalk correction
Band 21 MODIS scene shows less striping with crosstalk correction applied
Impact to L1B Data Set
Non-uniform channel to channel response of 0.5% to 1.0% in radiance has been reported.
Estimated Impact to L2 Science Products
The effect could introduce errors or uncertainties of up of 2 degrees Celsius in a 4-µm based Sea-Surface Temperature product.
Cause of Effect
The effect is due to the use of a linear calibration for these bands in the initial version of the algorithm, when the actual functioning of the detectors is non-linear (quadratic). Slight variations in the non-linear behavior of one detector to another will lead to an apparent striping in the radiance product for temperatures different than the temperature of the OBC-BB.
This situation is the result of the pre-launch electronics change for electronic cross-talk.
Mitigation Approach/Added Testing to Characterize Effect
On-orbit observations of an OBC-BB over the temperature range 274 K to 315 K enabled a determination of these non-linear terms, and these values were installed with the LUT update known as the St. Patrick’s Day update, beginning on 17 March 2000.
Estimated Schedule to Complete
History of Characteristic
|Epoch Number||Time Span||Version Number at start of epoch||General Characteristics||Improvements||Documentation|
|1||Feb. 24, 2000 - March 16, 2000 2000055.1550 - 2000076.2122||220.127.116.11||Imprecise calibration due to use of pre-launch gains.|
|2||March 16, 2000 - present 2000076.2122 - present||18.104.22.168||Updated gains from on-orbit study fixes the problem.||MCST library document M0973|