Infrared Band Destriping Software Package v1.3 release 7 November 2016
Software package which removes some of the Aqua and Terra
striping that appears in infrared bands as a result of detector-to-detector
variations and mirror side differences.
This algorithm has been shown to
have a positive impact on imagery as well as the cloud mask, cloud phase,
cloud top property, atmospheric profiles and sea surface temperature
products. It has been running operationally at NASA MODAPS
prior to implementation of the MODIS atmosphere team science software
since 2005. The software includes executables as well as a run script that chooses the
best coefficient file to use based upon the satellite platform (Aqua or Terra) and
the date/time of the input data.
The software utilizes the algorithm described in:
Weinreb et al., 1989:"Destriping GOES Images by
Matching Empirical Distribution Functions". Remote Sens. Environ., 29,
Changes in Version 1.3
Updates were made to the coefficient files and to the main
run script. The executables and source code have not changed.
The main change is a new coefficient file for Terra that helps to
improve the detector-to-detector consistency in the MODIS infrared bands
after the safe hold event that occurred in February 2016. MODIS bands
27-30 were impacted most by this event, and the new Terra .v5 coefficient file
Terra and Aqua MODIS Implementation in FORTRAN-90
- MODIS is treated as a 20 detector instrument in the thermal emissive bands
(10 detectors on each mirror side).
- The empirical distribution function (EDF) is computed for each detector
(cumulative histogram of relative frequency).
- The EDF for each detector is adjusted to match the EDF of a reference
- Algorithm operates on L1B scaled integers (0-32767).
- Median scaled integer value for each band is restored following destriping.
Impact on the L1B 1KM data
- Correction LUT is created for each individual granule on the fly.
- Uncorrected scaled integers are replaced with corrected scaled integers.
- Non-functional and very noisy detectors in some bands are replaced with neighbors.
- Requires less than 60 seconds to run for each granule.
- Text configuration file (one each for Terra and Aqua) defines:
- (a) which bands will be destriped
- (b) which detector will be the reference for each band
- (c) which detectors will be replaced for each band
An example of the effects of the V1.3 destriping software as executed
on two recent MODIS L1B files can be seen here.
- For most of the thermal infrared bands, the impact is positive. Striping noise
is significantly reduced.
- NOTE: The L1B 1KM file is irreversibly changed by this algorithm. It is
possible, but complicated, to implement this algorithm in such a way that the
destriping is reversible.