| 10 Jan 2013 09:20 AM |
|
Hey everyone,
I would like to have a look at a vegetation index (Vegetation Condition Index – VCI), globally available through NOAA and its Center for Satellite Applications and Research (STAR) ( http://www.star.nesdis.no...b/vci/VH/vh_ftp.php) . The data itself I downloaded from here: ftp://ftp.star.nesdis.noaa.gov/pub/corp/scsb/wguo/GVIx/GVIx_VH_16km/VCI/
The data is provided in a .hdf format. I have managed to load the data into ENVI 5.0 but now have difficulties to add the provided projection to the data itself.
From the metadata I know that the projection is Plate carée, this is the information provided:
Attribute #1: "FILENAME" "GVIX_NC_G16_C07_VCI_Y1984_P01.hdf"
Attribute #2: "SATELLITE" "NC"
Attribute #3: "PROJECTION" "Plate_Carree"
Attribute #4: "GRID_ROWS" 904
Attribute #5: "GRID_COLUMNS" 2500
Attribute #6: "START_LATITUDE_RANGE" 75.02400208
Attribute #7: "END_LATITUDE_RANGE" -55.15200043
Attribute #8: "START_LONGITUDE_RANGE" -180.00000000
Attribute #9: "END_LONGITUDE_RANGE" 180.00000000
Attribute #10: "SD name convention" "TYPE_YYYY_MM TYPE: Data Type, such as 'VCI' YYYY: year; four digits WW: period, two digits “
Attribute #11: "Input Files" "dirVH/GVIx_VH_G16/SMN/GVIX_NC_G16_C07_SMN_Y1984_P01.hdf
dirVH/GVIx_VH_G16/MXN/GVIX_G16_C07_MXN_P01.hdf dirVH/GVIx_VH_G16/MNN/GVIX_G16_C07_MNN_P01.hdf "
In ENVI the corresponding projection exists just under a different name (Equirectangular) and I am supposed to give information about:
- Sphere radius
- False easting
- False northing
- Latitude of true scale
- Longitude of central meridian
I do not know where I can find that information or how to convert the information provided. It is probably a not very difficult, but as I have tried all day, I thought posting it here might be a better way to solve my problem quickly. Any ideas? Or any other threads dealing with this problem?
Any help highly appreciated
Stef |
|
|
|
MariM
Veteran Member
Posts:2396  
| 10 Jan 2013 10:08 AM |
|
|
What have you tried? The equirectangular is already based on a sphere radius so you should not need to change that. For latitude of true scale and longitude of central meridian you can use to center the projection where the image is located. I would use the approximate center of your latitudes and longitudes provided in the information in your HDF.
|
|
|
|
Deleted User
New Member
Posts: 
| 11 Jan 2013 02:21 AM |
|
|
Hey MarM,
many thanks for your quick reply!!!
I have tried to typ in different values for the parameters required. I left sphere radus at the default value of 6370997.00 and changed lat and lon to the values of the center of the dataset. As it is a global dataset lon is 0 and lat is 9.936001 for the central point, as not all of the areas in North and the South are covered. I did this with the 32-bit version of ENVI so I could still use the version. This is how the dialog box looks like, but I do not know what to enter
Then I tried to work with the proper ENVI 5.0, but I realized, that in the new version they suddenly introduces geographic lookup tables (GLT) and IGM (do not even know what that means)....and now I am totally lost...
"Just" want to georefence a .hdf file using the information present in the Metadata(e.g. Plate Carree projection, start latitude range, end latitude range, start longitude range, end longitude range). I am sure someone has done it before....maybe there is an extension or an idl script, but have not found anything.....and manually I am unable to insert the information....
Still would appreciate any more help
Stella
|
|
|
|
Deleted User
New Member
Posts:
| 11 Jan 2013 09:41 AM |
|
|
Since you already have your Lat and Lon values (9.9 and 0) what happens if you enter those in the dialog? If you aren't sure about the procedure you can refer to the ENVI Help.
Also, I don't understand what you mean by "...new version they suddenly introduces geographic lookup tables (GLT) and IGM). If you could explain how those tables show up it can be helpful.
|
|
|
|
Deleted User
New Member
Posts:
| 05 Feb 2013 04:57 AM |
|
|
Hey,
sorry it took me some time to answer, but I still have not resolved the problem. The ENVIT help is not very specific. I have now loaded the hdf file into ENVI and tried to chage the map info in the header.
First I generated a new projection using equirectangular as projection type leaving the default value for the Sphere radius. I named the new projection Plate Carree. For all other values (false easting, false northing, latitutde of true scale and longitude of central meridian) I entered 0. This is how the dialog box appears:
I then started editing the Map Information the following way
I used the informatino profided in the the hdf file (shown in the first post), but must have done a mistake. Any suggestions on how to resolve the problem and project the data correctly?
Thankful for any help
Stef
|
|
|
|
MariM
Veteran Member
Posts:2396
| 05 Feb 2013 05:33 AM |
|
|
You should set your Latitude of true scale and Longitude of central meridian to be approximately where your image is centered. If it is a world image, then you can use 0 lat and 0 lon. What do you mean by it doesn't work? Where did you derive your pixel size? I think the parameters of the projection are ok but if your pixel size is off, then the coordinates will be correct in the upper left corner but get progressively off as you move towards the lower right. I think your pixel size may be 16800 and not 16000.
|
|
|
|
Deleted User
New Member
Posts:
| 06 Feb 2013 12:42 AM |
|
|
Hey,
thanks again for your reply. It is a world image and I have entered 0 lat and 0 lon, but when I load a different image into ENVI a major shift occurs (the two images do not even overlap). I also tried reprojecting the image to lat lon or to define a projection in ArcGIS. None of the things really worked, once I compare the data with other images. The dataset is supposed to have a spatial resolution of 16km, which is stated in the file name (G16 for ground resolution) as well as in ancillary information I have.
Greetings
Stef
|
|
|
|
Deleted User
New Member
Posts:
| 06 Feb 2013 03:43 AM |
|
|
Hey,
I am under the impression that it has something to do with the tie point matching. I thought, the X and Y values of the Image Coordinates of Tie Point referred to the lower left corner, but read somewhere (cannot remember where, but not the ENVI help!) that it refers to the upper left corner....so I tried to change the parameters accordingly:
But it did not work. When I chose the middle of the image as tie point it worked (just a little fine tuning needed). These are the parameters I entered:
Now I am wondering: what is meant by the x and y values of the Image Coordinates of Tie Point: the upper left or upper right corner of the image?
And I have new problems:
Managed to enter the appropriate values into the header file and export the whole thing as an geotiff. When I open it in ENVI it is fine, but once I open it in Arcmap, the projection is gone!!!
In addition I cannot project the file using the project raster tool from ENVI.
And then to my utter bewilderment I saw that with the project raster tool under predefined coordinate systems, there actually is a "Plate Carree" Projection, which is not there when I want to chose the appropriate projection in the header file. So when I want to reproject an image it is possible to chose this projection, but not when one wants to define one.....seems weird to me:
Any explanation for that?
Happy about any help.
Greetings
Stef
|
|
|
|
MariM
Veteran Member
Posts:2396
| 06 Feb 2013 03:37 PM |
|
|
It doesn't really matter which tie point you choose as long as the coordinate going along with it is appropriate for that pixel location.
I think the problem with the dataset is that when you define the projection, there is no datum and the Esri projection engine assumes a datum based on a sphere. The projection string produced from the projection you are defining is missing this information.
I would be curious to know if the data files should actually have a datum. I downloaded one of the datasets and I can't find anything in the metadata about a datum but the Plate Carree is often associated with a WGS84 datum. It would be good to confirm this.
For now, I would open the hdf image in ENVI with the projection you have defined and save it to an ENVI format file. This will copy the header information into the new file. Then, in this file, replace the coordinate system string with this:
coordinate system string = {PROJCS["World_Plate_Carree",GEOGCS["GCS_WGS_1984",DATUM["D_WGS_1984",SPHEROID["WGS_1984",6378137.0,298.257223563]],PRIMEM["Greenwich",0.0],UNIT["Degree",0.0174532925199433]],PROJECTION["Plate_Carree"],PARAMETER["False_Easting",0.0],PARAMETER["False_Northing",0.0],PARAMETER["Central_Meridian",0.0],UNIT["Meter",1.0]]}
This is one of the Esri projection strings based on WGS84. This provided fairly close coordinates but when I use a pixel size of 16.8km (16800), the coordinates were even closer.
|
|
|
|
Deleted User
New Member
Posts:
| 14 Feb 2013 08:31 AM |
|
|
Thanks MariM,
it seems to have worked. Just need the information from the data provider about the projection. Still waiting for it.
Greetings
Stef
|
|
|
|
Deleted User
New Member
Posts:
| 10 Apr 2013 12:53 PM |
|
|
test
|
|
|
|