| Re: AVHRR Image Mapping Problem [message #51842] |
Fri, 15 December 2006 10:17  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Jean H. writes:
> The data is correct... using ArcMap, the data lines up almost perfectly
> (i.e. a few places are off, but most pixels are within the borders)
I'm beginning to wonder if I have stumbled into the problem
that caused the ENVI developers to write their own map
projections, instead of using IDL's. :-(
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
|
|
|
|
| Re: AVHRR Image Mapping Problem [message #51845 is a reply to message #51844] |
Fri, 15 December 2006 09:44  |
Braedley
Messages: 57
Registered: September 2006
|
Member |
|
|
> From the image, it looks like a possible off by one error, maybe in the
scaling of the GeoTiff data?
The points furthest from the center appear to have the most error, and
they are shifted away from the center point. For example, if you look
along the French Coast, the error doesn't appear to be as large because
the direction of the error is close to the line that the coast makes.
Braedley
David Fanning wrote:
> kuyper@wizard.net writes:
>
>> One thing you could check is the possibility that the standard
>> parallels are reversed: does it work better if you use -21 and +19?
>
> Humm. Well, now the upper part of the map is correct, but
> the error is transferred to the lower part of the map.
> (I just multiplied the standard parallels by minus 1.)
> Does this give you any clues?
>
> Cheers,
>
> David
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming: http://www.dfanning.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: AVHRR Image Mapping Problem [message #51846 is a reply to message #51844] |
Fri, 15 December 2006 09:37 |
Matt[1]
Messages: 23
Registered: December 2006
|
Junior Member |
|
|
It's been my experience that it's never a good idea to check projection
accuracy using country outlines. ESRI shapefiles are notoriously
inaccurate when it comes to geolocation. I suspect that similar vector
files from other sources will have similar shortcomings, with few
exceptions. I actually know for a fact that National Geographic
sometimes resorts to stretching or compressing their raster data (using
graphics software - Photoshop perhaps?) to match their vector data for
their publication maps. What is the source of the country outline you
are using as comparison? Is it a vector file? If it is a vector file,
could you search for a more accurate one?
If you can, I would suggest using SRTM elevation data
(http://glcf.umiacs.umd.edu/data/srtm/, access:
http://glcfapp.umiacs.umd.edu:8080/esdi/index.jsp) to check projection
accuracy. As far as I know SRTM data has a high degree of geolocational
accuracy. Use the SRTM 30 Arc Second, which is also at 1km. Keep in
mind that SRTM data is not in Albers projection so you'd have to warp
the data. I imagine that this will lead to some misalignments since you
are warping SRTM from Geographic coordinates to Albers and the AVHRR
data was warped from Goodes Interrupted to Albers, but it should give
you a better idea of the relative geolocational accuracy of the two
datasets.
If you'd rather use vector data to do the comparison, try the SRTM
Water Bodies data http://edc.usgs.gov/products/elevation/swbd.html.
This is a very accurate vector file outlining coastlines/waterbodies.
Best
Matt S.
On Dec 15, 11:11 am, Paul van Delst <Paul.vanDe...@noaa.gov> wrote:
> David Fanning wrote:
>> Folks,
>
>> Does anyone have any experience working with AVHRR NDVI
>> image data or Albers map projection? I have obtained
>> the data, which is of the African continent from here:
>
>> ftp://ftp.glcf.umiacs.umd.edu/glcf/GIMMS/Regional/Albers/Afr ica
>
>> The image is in an Albers Conical equal area projection
>> and the centers of the four corner pixels are known from
>> the documentation:
>
>> ; YX coordinates of the four corners (LL, UL, UR, LR)
>> longitude = [-23.49, -24.6, 64.523, 63.414]
>> latitude = [-42.243, 43.711, 43.712, -42.242]
>
>> This is a GeoTiff file, so I also pull the Standard
>> Parallels out of the geotiff information stored in
>> the file (they are -19 and 21).
>
>> I follow the method outlined on this page (which has
>> worked perfectly for a polar stereo map projection),
>> using instead of a Stereo projection, an Albers
>> projection with standard parallels:
>
>> http://www.dfanning.com/map_tips/precipmap.html
>
>> The method *ALMOST* works! :-)
>
>> But the continental outlines do not QUITE line up properly.
>> You can see my result here:
>
>> http://www.dfanning.com/misc/africa.jpg
>
>> Do you think this might be an Albers projection problem?
>> A difference between MAP_PROJ_INIT and MAP_SET? (I have
>> tried different DATUMS with no change in effect.)
>
>> Or, do you think this might just be right? :-(Not my area of expertise, but it sure looks like some sort of projection problem. If it
> was an issue with the data (e.g. AVHRR geolocate issue) then I think it would be shifted
> in one direction everywhere. Your test plot shows a eastward shift on the northern east
> coasts, a westward shift on the northern west coast, and a much smaller westward shift on
> the southern west coast. That suggests to me the data near your standard parallels are
> more "accurate" (by whatever measure) but things get smudged out more and more as you move
> away from them.
>
> paulv
>
> --
> Paul van Delst Ride lots.
> CIMSS @ NOAA/NCEP/EMC Eddy Merckx
> Ph: (301)763-8000 x7748
> Fax:(301)763-8545
|
|
|
|
|
|
| Re: AVHRR Image Mapping Problem [message #51848 is a reply to message #51847] |
Fri, 15 December 2006 09:21 |
James Kuyper
Messages: 425
Registered: March 2000
|
Senior Member |
|
|
David Fanning wrote:
> Folks,
>
> Does anyone have any experience working with AVHRR NDVI
> image data or Albers map projection? I have obtained
> the data, which is of the African continent from here:
>
> ftp://ftp.glcf.umiacs.umd.edu/glcf/GIMMS/Regional/Albers/Afr ica
>
> The image is in an Albers Conical equal area projection
> and the centers of the four corner pixels are known from
> the documentation:
>
> ; YX coordinates of the four corners (LL, UL, UR, LR)
> longitude = [-23.49, -24.6, 64.523, 63.414]
> latitude = [-42.243, 43.711, 43.712, -42.242]
>
> This is a GeoTiff file, so I also pull the Standard
> Parallels out of the geotiff information stored in
> the file (they are -19 and 21).
One thing you could check is the possibility that the standard
parallels are reversed: does it work better if you use -21 and +19?
|
|
|
|
| Re: AVHRR Image Mapping Problem [message #51849 is a reply to message #51848] |
Fri, 15 December 2006 08:11 |
Paul Van Delst[1]
Messages: 1157
Registered: April 2002
|
Senior Member |
|
|
David Fanning wrote:
> Folks,
>
> Does anyone have any experience working with AVHRR NDVI
> image data or Albers map projection? I have obtained
> the data, which is of the African continent from here:
>
> ftp://ftp.glcf.umiacs.umd.edu/glcf/GIMMS/Regional/Albers/Afr ica
>
> The image is in an Albers Conical equal area projection
> and the centers of the four corner pixels are known from
> the documentation:
>
> ; YX coordinates of the four corners (LL, UL, UR, LR)
> longitude = [-23.49, -24.6, 64.523, 63.414]
> latitude = [-42.243, 43.711, 43.712, -42.242]
>
> This is a GeoTiff file, so I also pull the Standard
> Parallels out of the geotiff information stored in
> the file (they are -19 and 21).
>
> I follow the method outlined on this page (which has
> worked perfectly for a polar stereo map projection),
> using instead of a Stereo projection, an Albers
> projection with standard parallels:
>
> http://www.dfanning.com/map_tips/precipmap.html
>
> The method *ALMOST* works! :-)
>
> But the continental outlines do not QUITE line up properly.
> You can see my result here:
>
> http://www.dfanning.com/misc/africa.jpg
>
> Do you think this might be an Albers projection problem?
> A difference between MAP_PROJ_INIT and MAP_SET? (I have
> tried different DATUMS with no change in effect.)
>
> Or, do you think this might just be right? :-(
Not my area of expertise, but it sure looks like some sort of projection problem. If it
was an issue with the data (e.g. AVHRR geolocate issue) then I think it would be shifted
in one direction everywhere. Your test plot shows a eastward shift on the northern east
coasts, a westward shift on the northern west coast, and a much smaller westward shift on
the southern west coast. That suggests to me the data near your standard parallels are
more "accurate" (by whatever measure) but things get smudged out more and more as you move
away from them.
paulv
--
Paul van Delst Ride lots.
CIMSS @ NOAA/NCEP/EMC Eddy Merckx
Ph: (301)763-8000 x7748
Fax:(301)763-8545
|
|
|
|
|
|
|
|
| Re: AVHRR Image Mapping Problem [message #51991 is a reply to message #51847] |
Fri, 15 December 2006 10:23 |
James Kuyper
Messages: 425
Registered: March 2000
|
Senior Member |
|
|
David Fanning wrote:
> kuyper@wizard.net writes:
>
>> One thing you could check is the possibility that the standard
>> parallels are reversed: does it work better if you use -21 and +19?
>
> Humm. Well, now the upper part of the map is correct, but
> the error is transferred to the lower part of the map.
> (I just multiplied the standard parallels by minus 1.)
> Does this give you any clues?
Actually, yes. I have no idea why the standard parrallels might be
incorrect, but given the symptoms you describe, it might be worthwile
to try 19.99999 and -20.00001. IDL doesn't like it when the standard
parallels add up to exactly zero. That is unfortunate, because a Albers
projection with standard parallels adding up to zero is a different way
of describing the cylindrical equal-area projection, one of my favorite
map projections, and one that IDL doesn't support.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
