Aqua MODIS Calcite product

[larteaga]

Hi all,
There seems to be an issue with the PIC (SMI) product (Balch and Gordon) for monthly 9km files, after July 2014. I only find negative (<0) retrievals for all global grids from this period on. I have also downloaded monthly files for 2013, and those seem to work fine. Is there a problem with the algorithm after July, 2014?

Cheers,
Lionel

[OB WebDev - norman]

Hi Lionel,

Are you applying the scale (slope) and offset (intercept) to the data?
Compare, for example, August 2013 with August 2014.

h5dump -a /pic/scale_factor A20132132013243.L3m_MO_PIC_pic_9km.nc
HDF5 "A20132132013243.L3m_MO_PIC_pic_9km.nc" {
ATTRIBUTE "scale_factor" {
   DATATYPE  H5T_IEEE_F32LE
   DATASPACE  SIMPLE { ( 1 ) / ( 1 ) }
   DATA {
   (0): 1
   }
}
}

h5dump -a /pic/add_offset A20132132013243.L3m_MO_PIC_pic_9km.nc
HDF5 "A20132132013243.L3m_MO_PIC_pic_9km.nc" {
ATTRIBUTE "add_offset" {
   DATATYPE  H5T_IEEE_F32LE
   DATASPACE  SIMPLE { ( 1 ) / ( 1 ) }
   DATA {
   (0): 0
   }
}
}

h5dump -a /pic/scale_factor A20142132014243.L3m_MO_PIC_pic_9km.nc
HDF5 "A20142132014243.L3m_MO_PIC_pic_9km.nc" {
ATTRIBUTE "scale_factor" {
   DATATYPE  H5T_IEEE_F32LE
   DATASPACE  SIMPLE { ( 1 ) / ( 1 ) }
   DATA {
   (0): 2e-06
   }
}
}

h5dump -a /pic/add_offset A20142132014243.L3m_MO_PIC_pic_9km.nc
HDF5 "A20142132014243.L3m_MO_PIC_pic_9km.nc" {
ATTRIBUTE "add_offset" {
   DATATYPE  H5T_IEEE_F32LE
   DATASPACE  SIMPLE { ( 1 ) / ( 1 ) }
   DATA {
   (0): 0.065
   }
}
}

Regards,
Norman

[larteaga]

Hi Norman,

Many thanks for your reply. I was not aware that these two parameters needed to be applied. This seems to solve the problem.
As a follow up, Is there a reason why they are not automatically applied to the data?

Cheers,
Lionel

[OB WebDev - norman]

Hi Lionel,

I believe that SeaDAS should automatically apply the scaling.
Other packages would not necessarily be coded to do that.

Norman

[gnwiii]

You should mention what software you are using.   Current NetCDF4-CF libraries should handle scaling and missing value coding "automagically".  Older libraries often manual intervention.  You should avoid using hard-coded values in your own software, as the values may change.

[larteaga]

I see. I use mostly MATLAB, and thus, I write all the processing code manually. I don't know how relevant this is, but I initially wanted to use the BIN files (of the same PIC product), and was not able to open them with MATLAB's Netcdf library. Have you got any thoughts on that?

Cheers,
Lionel

[gnwiii]

You might get better answers posting your MATLAB code in the "Non-SeaDAS Packages (e.g. MATLAB, ENVI, GIS, etc)" Section.   You should mention your MATLAB version and the error you get when you try to open a binned file.  NetCDF4 actually uses the HDF5 library, so Using NASA NetCDF binned data with MATLAB's HDF5 functions uses hdf5 to get around problems with unsupported data types in MATLAB's ncread.   NetCDF-4 (Network Common Data Form, version 4) has some notes on support for NetCDF4-CF in older software.  I have also had good results with Unidata's NetCDF-Java in MATLAB and also R.

[larteaga]

Ok, many thanks for your prompt help!
Cheers,
Lionel