What is the status of the data quality in the red bands, NIR, and SWIR for MODIS-Terra? I understand that the blue and green (and band ratios using those bands), can be unreliable from Terra based on the information presented in Franz et al. (2008). I am using the 250m spatial resolution band 645nm from MODIS-Aqua and hopefully MODIS-Terra. I see some striping in the L2 Terra scenes that I have generated locally at the higher spatial resolution with SeaDAS 7.5.3 on mac. I am specifically looking at nLw_645 and Rrs_645.
If Terra is useful for 645 nm, is there a recommended method for combining Aqua and Terra nLw_645 and Rrs_645 at the 250m spatial resolution? For example, binning into averaged daily, weekly, and monthly composites with attention to the differences between to two MODIS sensors? Any advice would be much appreciated.
It is true that MODIS-Terra has issues with the blue bands. we have for a number of years employed a cross-calibration to the SeaWiFS and MODIS-Aqua sensors to help mitigate these errors (https://doi.org/10.1364/AO.47.006796). While we are able to process the 645nm band, it was not intended for use in ocean color applications. The bandpass is quite broad and the SNR rather low relative to the typical ocean color band set. This is not to say it is not useful, but just a caveat to keep in mind. The SWIR bands on MOIDS, however, have even lower SNR than the 645nm band, so much so that we do not bother to vicariously calibrate them. Their utility for ocean color is quite low.
Despite our calibration efforts, detector stripes and mirror side banding will always be visible at some level. You would need to decide for yourself if those residual calibration artifacts are at an acceptable level for your purpose. The data may be suitable for qualitative analyses or feature tracking, but likely less so for long-term climate analyses.
If you desire to combine MODIS-Terra and MODIS-Aqua, the tool to use would be
l2bin. You can use
l2binto generate daily or even multi-day composites, but if your goal is multi-day, I would bin to dailies with l2bin and combine the dailies into multi-day composites using
l3bin. As the output bin file is not in a raster format, you may also want to use
Hope this helps,
That all makes sense! Thanks for the reference. Good to know you would recommend binning Aqua and Terra together to daily before doing weekly/monthly, that is especially helpful.