This characteristic of sensor behavior is called sensor Response versus Scan Angle (RVS). In the MODIS Earth View sector (the portion of the scan pointing toward Earth in normal operation), the scan angle varies from -55 degrees at beginning of scan to +55 degrees at end of scan
Impact to L1B Data Set
We anticipate that there will be an effective mirror side (10 km at Nadir) striping in the data sets for Bands 27 to 36 due to mirror side differences in RVS. Stripes in the L1B or L2 products appear from one scan line (10 km along-track displacement at Nadir) to the following scan line. These stripes are imperceptible at the scan angle about midway between beginning of scan and Nadir, but become more apparent as the scan angle deviates from this location. These RVS errors also cause asymmetry in the across-track radiance profiles over flat scenes for all channels.
Estimated Impact to L2 Science Products
RVS errors in Bands 31 and 32 are believed to be large enough so that the L1B product will not meet the measurement requirements. Product accuracy is a function of mirror side and scan angle position within MODIS field of view; would require validation studies to track mirror side and scan angle position to assess quality of L2 products. Sea-Surface Temperature investigators and atmospheric sounding investigators (covering bands 31 - 36) have reported such effects in the L1B product, and they have not yet been fully studied.
Cause of Effect
The effect is the result of mirror side differences in RVS. The pre-launch system Response Versus Scan angle (RVS) for the Terra MODIS was not determined. Consequently, the RVS used initially for Terra MODIS was constructed from the Aqua MODIS measurement for RVS combined with a National Physical Laboratory measurement of the reflectance of witness samples corresponding to the two sides of the Terra MODIS scan mirror. MCST averaged the two mirror side RVS values and used this average for each mirror side.
On-orbit calibration in these bands is performed using the On-board Calibrator Blackbody (OBC-BB) source, which is located at a scan mirror angle of incidence (AOI) of 26.7 degrees. This AOI corresponds to a scan angle nearly half-way between the beginning of scan and Nadir. This approach provided an RVS that led to significant differences in the resulting data product, in 10-km wide swaths, for scan angles away from the angle of the OBC-BB. At the angle of the OBC-BB, the differences in mirror side RVS are captured in the linear gain coefficient.
The RVS effect is caused by the specular reflectance and emittance of the two-sided scan as a function of AOI mirror. The polarization of light reflected from the scan mirror also changes with the AOI onto the scan mirror. The RVS for Bands 20 - 25, (3.7 to 4.5 µm) varies by only a small amount. The RVS for bands 27 - 30 (6.6 to 8.7 µm) is expected to vary up to 7% and there is not presently an effective measurement for this quantity. The RVS for Bands 31 - 36 (10.5 to 14.5 µm) is expected to vary up to 2.5% and there is not presently an effective measurement for this quantity.
Mitigation Approach/ Added Testing to Characterize Effect
A change was installed in production software to improve the handling of certain RVS terms in the IR. This improved RVS was based on analysis of the backside of the closed nadir aperture door data (April 26-27, 2000). In this test, the RVS on mirror side one was adjusted so that the observed radiance for mirror side one matched that observed for mirror side two. This improved RVS value was installed in LUT V220.127.116.11 on August 10, 2000 to eliminate mirror side differences by detector. The change was made only for mirror side 2.
Further improvements await the deep space calibration maneuver.
Estimated Schedule to Complete
Data for establishing the RVS necessary to meet SST measurement accuracy requirements can be obtained only through a spacecraft maneuver to look at cold dark space for at least 15 minutes in duration. The maneuver is not expected to be performed until late summer of 2002 at the earliest. The reasoning for this delay was to allow the EOS Operations Center time to adjust to operating the Terra and Aqua spacecraft simultaneously.
Images comparing data before and after the August 10, 2000 change
Before August 10, 2000:
After August 10, 2000:
History of Characteristic
|Time Span||Version Number at|
start of epoch
|1||Feb. 24, 2000 - June 10, 2000|
2000055.1550 - 2000162.1155
|18.104.22.168||RVS was determined from scan mirror (witness sample) reflectance (NPL) and FM1 fixed optics polarization parameters with no mirror side and detector dependence. Code had an index bug such that an incorrect RVS correction from the scan mirror was introduced. The effect was more obvious in PC bands. Another bug existed in the preprocess that affected the PC crosstalk correction||Not applicable.||MCST Memo on May 13, 1999|
|2||June 10, 2000 - Aug. 18, 2000|
2000162.1155 - 2000231.1320
|22.214.171.124||Same pre-launch determined RVS with no mirror side and detector dependence. Code index bug related to incorrect RVS correction from the scan mirror was fixed. The bug in the preprocess that affected the PC crosstalk correction still present.||Signicant improvement of TEB radiance product; effective reduction of radiance at end of scan||MCST Memos on Aug. 02, 2000 and on Aug. 03, 2000|
|3||Aug. 18, 2000 - ..|
2000231.1320 - ..
|126.96.36.199||Mirror side and detector dependent relative RVS was determined from on-orbit closed nadir door data. The bug in the preprocess was fixed.||Reduced 10- km mirror side-induced "banding" at all scan angles away from angle at OBC-BB, but no effective improvement in nature of overall knowledge of RVS.||MCST memos on July 29, 2000 and on Aug. 25, 2000|