Page 1 of 1

l2gen resolution 250 m

Posted: Thu Jul 27, 2017 4:39 am America/New_York
by cassandranormandin
Hello,

I am actually working with MODIS Terra/Aqua L1B, created with SeaDAS. I've downloaded L1A on your website, then used modis_GEO (to create geofile), modis_L1B (to create files at 1KM, 500m and 250m) and then l2gen.
I want to apply atmospheric corrections on these images using l2gen. I've selected different parameters and I would like to have a spatial resolution of 250 m.

# PRIMARY INPUT OUTPUT FIELDS
ifile=/home/cnormandin/Documents/MODIS_images/MOD021KM.A2016015.1020.006.2016015200305.hdf
geofile=/home/cnormandin/Documents/MODIS_images/MOD03.A2016015.1020.006.2016015170142.hdf
ofile=/home/cnormandin/Documents/MODIS_images/T2016015102000.L2_LAC_OC

# SUITE
suite=OC

# PRODUCTS
l2prod=Rrs_nnn

# PROCESSING OPTIONS
aer_opt=-1
maskhilt=0
maskland=0

# SUBSETTING OPTIONS
north=47
south=43.5
east=1
west=-3

# THRESHOLDS
cloud_thresh=0.018
cloud_wave=2130

# ANCILLARY INPUTS  Default = climatology (select 'Get Ancillary' to download ancillary files)

# MISCELLANEOUS
resolution=250

It works well except the fact that the spatial resolution seems to be 1 km and not 250 m, although I've selected 250 m.
I don't understand why.

Thank you in advance for your help,
Cassandra

l2gen resolution 250 m

Posted: Thu Jul 27, 2017 7:47 am America/New_York
by OB.DAAC - SeanBailey
Casandra,

The 1km resolution bands will not magically obtain a higher resolution when processed at 250m.  The only bands that will truly be at 250m are the 645 and 859nm bands.
The others will interpolated (but effectively be supersampled) to 250m resolution and will appear blockier than the native 250m bands as a result.

Sean

l2gen resolution 250 m

Posted: Thu Jul 27, 2017 8:53 am America/New_York
by cassandranormandin
Thank you Sean!
But I don't understand why my bands at 250 m appear with a spatial resolution of 1 km too.. Like the others!
I've joined the file, it is corresponding to the 859 nm band. You can see that one pixel have a size of 0.01° (1km).

Cassandra

l2gen resolution 250 m

Posted: Thu Jul 27, 2017 11:13 am America/New_York
by OB SeaDAS - knowles
Hello Cassandra,

I don't quite follow this logic and where modis_L1B comes into play as you've described it. 

On a side note and just to be clear "our" site is :
https://oceancolor.gsfc.nasa.gov/cgi/browse.pl
or specific to your time of your file: https://oceandata.sci.gsfc.nasa.gov/MODIS-Aqua/L1/2016/015/

Our level-1A file is: A2016015102000.L1A_LAC
Then running modis_L1B creates A2016015102000.L1B_LAC
Then running modis_GEO creates A2016015102000.GEO
Then ifile=A2016015102000.L1B_LAC, geofile=A2016015102000.GEO would be used for running l2gen.

You might consider trying obtaining our file from our site (note: it is the same source data scene, but it is not the same file) and then comparing the results with what you are seeing when using the MOD files.

Danny

l2gen resolution 250 m

Posted: Thu Jul 27, 2017 11:52 am America/New_York
by cassandranormandin
Hello Danny,

Thank you for your reply.
I've downloaded L1A from your site https://oceandata.sci.gsfc.nasa.gov/MODIS-Aqua/L1/2016/015/
In fact to do modis_L1B, we need geofile, that's why I did modis_GEO before modis_L1B.
I could not do modis_L1B without geofile...

Ok I will compare this!

Cassandra

l2gen resolution 250 m

Posted: Thu Jul 27, 2017 12:30 pm America/New_York
by OB SeaDAS - knowles
Yes, you are right, the modis_GEO step I mentioned does need to occur before the modis_L1B step.

I noticed that you are using climatology default for your ancillary.  In order to get better atmospheric correction data specific to your scene and time you should be running "Get_Ancillary" in the l2gen GUI, or from command line run getanc.py and put the results into a par file which you add to the l2gen run as an option (par=myAncFile.par).

Danny

l2gen resolution 250 m

Posted: Thu Jul 27, 2017 6:57 pm America/New_York
by OB.DAAC - SeanBailey
Cassandra,

The resolution is only a nominal resolution at nadir.  Off nadir, the pixels will grow.  For example, a 1km nominal pixel is more like 2x5 km at the edge of the swath.

Sean