Ancillary File Names
To accomodate a growing suite of required ancillary inputs, the ancillary file naming convention was standardized. The names will all be in the form Nyyyydddhh_qqq_sss_ppp.xxx where:
- xxx is the file-type suffix
- yyyy is the start year
- ddd is the start day
- hh is the optional start time in hours
- qqq is the quantity
- sss is the source
- ppp is the period of application
As an example, the full set of ancillary files required to process one SeaWiFS file (S1998199173926.L2_GAC_OC in this case) looks like this:
This list includes two unique meteorological (MET) files, two ozone (O3) files, a sea ice file, and an sst reference file.
Meteorological data changes
The new aerosol determination algorithm uses surface relative humidity as an ancillary quantity. An examination of the relative humidity from the NCEP meteorological data source (plot below, blue data points) revealed that there were several times that the global ocean average relative humidity had unusally large shifts in the average (slow 2% decrease in 2000 and abrupt 2% increase in 2005). Shifts in the ancillary data of this sort can result in shifts in the ocean color parameters retrieved, which would be scientifically unacceptable and would make it more difficult to make improvements in the other processing algorithms and calibration.
The NCEP/NCAR reanalysis (plot below, green data points) is complete and the relative humidity does not show shifts that are in the operational data. Unfortunately, this reanalysis has significantly lower resolution: 2.5 degree grid size as compared to the 1 degree grid for operational data.
The NCEP/NCAR reanalysis and the operational relative humidity were combined to make daily humidity fields with the temporal stability of the reanalysis but retain the higher resolution features of the operational data (plot below, red data points). This was achieved by correcting the operational fields by the difference between the reanalysis field and operational fields smoothed over a 10 degree region. Relative humidity fields used operationally will be derived only from the operational NCEP data. The stability of that data will be monitored and corrected with the reanalysis as needed to preserve a stable series.
This plot shows the average relative humidity over the global ocean from 1997 to the end of 2008 derived from 3 sources:
- The operational NCEP datasets (in blue)
- The NCEP / NCAR reanalysis (in green)
- The corrected data series to be used in the reanalysis (in red)
Ozone data changes
The total ozone that has been in use for the processing of ocean color data is derived from 2 main sources: EP TOMS data from 1997 to the end of 2005, and TOAST (a combination of SBUV and TOVS data) onward. Some problems in the temporal trend of ocean color retrievals at the time of the EPTOMS-TOAST transition has prompted us to search for a more consistent series of total ozone data for testing and ultimately reprocessing and operational use. We are replacing the TOAST source with OMI-derived total ozone fields. The TOAST will remain as a backup for use if gaps in the OMI of greater than 1 day occur.
In addition, a new ozone climatology has been developed from the EP-TOMS data for use should both OMI and TOAST be missing. This climatology is a daily climatology, with an average day of ozone for 366 days of the year.
New NO2 climatology
The new NO2 corrections require a time-series of tropospheric and stratospheric NO2 concentrations spanning the lifetime of the ocean color missions. To that end, NO2 data from GOME, SCIAMACHY, and OMI were combined and adjusted, as described in Robinson et al. to create a consistent ancillary NO2 dataset of the required temporal extent at the highest temporal resolution possible. In addition, a monthly climatology was constructed based on the multi-sensor record. Given the difficulties in constructing and maintaining the daily NO2 dataset, it was decided that the monthly climatology will be utilized for this reprocessing. However, the daily NO2 will be produced and made available for community evaluation and regional processing.
New sea ice data requirement
The sea ice flag is now derrived from NSIDC near real time ice fraction data. For more information see the sea ice documentation.
Revised SST reference
The MODIS SST algorithm requires a reference field for initial guess and quality assessment. The Reynolds optimally interpolated 1-deg grids on 8-day time-scales was the source for this reference SST field in past reprocessings. All processing will now use the new Reynolds 0.25-deg daily optimally interpolated SST fields. In the OBPG environment, these netCDF files are simply renamed for consistency with standard conventions. As in previous reprocessings, there is no sea surface temperature (SST) requirement for ocean color processing; however, it is expected that SST knowledge will be required for future ocean color product suites such as inherent optical properties.