Location not found in image

By daurin Date 2012-05-18 13:12

Hello,

I am running automated queries to oceancolor.gsfc.nasa.gov/cgi/browse.pl and my script found (unfortunately by crashing) that a queried location was not found in the image returned. The search was for July 25, 1998 at a location between 38.5975 and 38.6975 N, and between -76.5490 and -76.4490 E. One of the granules returned was S1998206160318.L2_MLAC_OC. My script tried to subset it to a 5x5 around the station location by first running:

lonlat2pixline -x 2 -y 2 /data/Imagery/GLOBAL/seawifs/L2/S1998206160318.L2_MLAC_OC -76.4990 38.6475

which returned null value. Upon closer inspection of the image, I find the closest the edge gets to my location is 38.6517N -75.4180, about 100 km away.

I just ran the same query manually on the website with the same result. Is it normal for the query to pull images that do not actually have the query location in them?

Thanks,
Dirk

By @norman Date 2012-05-18 13:45

Hi Dirk,

> I just ran the same query manually on
> the website with the same result. Is it
> normal for the query to pull images that
> do not actually have the query location
> in them?

Yes this happens occasionally. I haven't
checked your case specifically, but I assume
that it is similar to the case I referred to in
this posting.
http://oceancolor.gsfc.nasa.gov/forum/oceancolor/topic_show.pl?pid=12112#pid12112

See also the following from the browser help pages.
http://oceancolor.gsfc.nasa.gov/BROWSE_HELP/quadsphere.html

Regards,
Norman

By daurin Date 2012-05-18 14:37

I see. Nice graphics. Thanks.

By daurin Date 2012-05-23 14:00

Sorry to revive a dead thread, but I'm still confused about some of these match-ups. The explanations above relate to false positives around the edges of areas of interest depending on how the search is conducted. I seem to be finding "false positives" in the middle of granules.

Take the example of 18 Feb 2002 (day 11736 since 1/1/1970) at between -66.5100 and -66.4100 N, and between -67.7000 and -67.6000 E. The first attached image shows the spot on Google Earth. Searching for Aqua, Terra, and SeaWiFS LACs I get the following matches:

************TERRA on 11736 between -66.5100 and -66.4100 N, and between -67.7000 and -67.6000 E
Matches: T2002049122500.L2_LAC_OC T2002049140500.L2_LAC_OC
************SeaWiFS on 11736 between -66.5100 and -66.4100 N, and between -67.7000 and -67.6000 E
Matches: S2002049105913.L2_MLAC_OC S2002049123800.L2_MLAC_OC S2002049141648.L2_MLAC_OC S2002049155535.L2_MLAC_OC

The second attached image shows a manual search on the oceancolor website that corroborates this, and the thumbnails show my location to be in the near-center (east-west) of at least three of these. Nevertheless, running latlon2pixline on these yields no results:

lonlat2pixline -x 2 -y 2 T2002049122500.L2_LAC_OC -67.6500 -66.4600
Failed to find the pixels in this image
lonlat2pixline -x 2 -y 2 T2002049140500.L2_LAC_OC -67.6500 -66.4600
Failed to find the pixels in this image
lonlat2pixline -x 2 -y 2 S2002049105913.L2_MLAC_OC -67.6500 -66.4600
Failed to find the pixels in this image
lonlat2pixline -x 2 -y 2 S2002049123800.L2_MLAC_OC -67.6500 -66.4600
Failed to find the pixels in this image
lonlat2pixline -x 2 -y 2 S2002049141648.L2_MLAC_OC -67.6500 -66.4600
Failed to find the pixels in this image
lonlat2pixline -x 2 -y 2 S2002049155535.L2_MLAC_OC -67.6500 -66.4600
Failed to find the pixels in this image

So, I get six matched images, some of which clearly cover my location, and yet none yield a result in lonlat2pixline. Thoughts?

By @sean Date 2012-05-23 20:12

Dirk,

I've run your lonlat2pixline command on three systems with seadas6.3 installed....all return a result:
try using the -v option - lonlat2pixline will spit out more information that *may* help us figure out why
it's not working for you.

e.g. lonlat2pixline -x 2 -y 2 -v T2002049122500.L2_LAC_OC -67.6500 -66.460 Input file T2002049122500.L2_LAC_OC is HMODIST Level-2 Data. lonlat2pixline 100% done. Reading scan: 2000 out of 2029 #Lon=-67.668892 #Lat=-66.454529 sline=1515 eline=1519 spixl=136 epixl=140

Sean

By daurin Date 2012-05-24 13:42

Thanks for checking that Sean. I figured out that a large proportion of my lonlat2pixline errors were deriving from an inexplicable failure to bunzip2 properly. I've put a check in the script to ensure bunzip2 ran without error prior to calling lonlat2pixline, and that has considerably reduced the fail rate. I still get hundreds of fails (instead of thousands), but until I learn otherwise, I am going to assume they are the type of errors Norman was referring to.
Dirk