Greetings!
I'm working on turning data arrays into geoTiff files. Basically, I have 3 arrays: 1 for my actual data, and 1 for latitude and 1 longitude both corresponding to a single point in my data array. To do this, one of my procedures involves projecting my data array into, well, my desired projection. I found this code called imagemap by Liam. I used this as the basis for my code. Here's what I do:
;set/get size of size of desired output image
ns = !d.x_size
nl = !d.y_size
;create a base map projection
map_set, /mercator, limit = [ latmin, lonmin, latmax, lonmax ]
;convert my lat-lon variables from data space to normal space
p = convert_coord( lon, lat, /data, /to_normal)
;scale my data array and assign it to a new array "newimage". (newimage is the output)
newimage( p( 0, * ) * ( ns - 1 ), p( 1, * ) * ( nl - 1 ) ) = bytscl( image, min = imin, max = imax, top = !d.table_size - 2 ) + 1B
This code produce a projected image as desired. I open the images (and I also have a shape file for reference) in ENVI to check if my geotiff is in the right size, scale and place. Apparently, the output image I get is too small and is way off of my desired size and scale. Soooo, I set ns and nl to my desired size since !d.x_size and !d.y_size were too small. Here's what I did:
;Create GLT file from my lat-lon array
ENVI_DOIT, 'ENVI_GLT_DOIT', DIMS=dim, I_PROJ=i_proj, O_PROJ=i_proj, out_name='glt_150501', R_FID=glt_fid,rotation=0,X_FID=night_lon_id,X_POS=0,Y_FID=ni ght_lat_id,Y_POS=0
;Extract the number of samples and lines from the GLT
ENVI_FILE_QUERY,glt_fid,ns=ns,nl=nl
This enlarged my image to my desired size but the scale is still wrong. There seems to be a minor offset. My data doesn't reach the edges of my image as I'd expected. The !x.window and !y.window changes when I use map_set. They weren't set to the lowest and highest values of the normal space.
This is where I think the problem occurs. When I call convert_coord and check the p-variable, my max/min lat-lon values aren't set to 0/1 (like in normal space). Instead, they are set to:
!x.window(0) is 0.0166717
!x.window(1) is 0.983338
!y.window(0) is 0.274023
!y.window(1) is 0.945210
This lead me to think that !x and !y variables had something to do with it. The size of my desired output image is 4775x3525. Using the information from !x/y.window, I get 4626x2365. When I trim the offset, I get an image of 4617x3236. This is close to my calculated window sizes (at least for the number of samples).
So this is what I did:
;I set the values equal to the maximum values I can get from the normal space: 0 and 1
!x.window(0) = 0
!x.window(1) = 1
!y.window(0) = 0
!y.window(1) = 1
;Apparently this does nothing.
That did NOTHING. It had no effect on my code so I tried to make my own simple convert_coord substitute. (This is what my present code looks like right now:
;;;;;;;BEGIN;;;;;;;;
;Create GLT file from my lat-lon array
ENVI_DOIT, 'ENVI_GLT_DOIT', DIMS=dim, I_PROJ=i_proj, O_PROJ=i_proj, out_name='glt_150501', R_FID=glt_fid,rotation=0,X_FID=night_lon_id,X_POS=0,Y_FID=ni ght_lat_id,Y_POS=0
;Extract the number of samples and lines from the GLT (This turned out to have the exact size I wanted)
ENVI_FILE_QUERY,glt_fid,ns=ns,nl=nl
!x.window(0) = 0
!x.window(1) = 1
!y.window(0) = 0
!y.window(1) = 1
;create a base map projection
map_set, /mercator, limit = [ latmin, lonmin, latmax, lonmax ]
;;;;Convert_coord function attempt
;;;;Just a bunch of variables I used to hasten my conversion process
diffLon = lonmax - lonmin
diffLat = latmax - latmin
sizes = size(lon)
lenX = sizes(1)
lenY = sizes(2)
print,lenX,lenY
p = fltarr(3,lenX*lenY)
;;;;;loop that converts my lat-lon array to normal space and assigns to variable p
for i=0, lenX-1 do begin
for j=0, lenY-1 do begin
;formula for properly assigning my the correct index to p
index = i*lenY + j
; This was the scaling formula I used to convert my data space to normal space
p(0,index) = (lon(i,j) - lonmin) / diffLon
p(1,index) = (lat(i,j) - latmin) / diffLat
p(2,index) = 0
endfor
endfor
;scale my data array and assign it to a new array "newimage"
newimage( p( 0, * ) * ( ns - 1 ), p( 1, * ) * ( nl - 1 ) ) = bytscl( image, min = imin, max = imax, top = !d.table_size - 2 ) + 1B
So what happened? I got my data to fit the extents of my image like I wanted BUT my data looks like a bunch of static. Any tips on how to work around my problem?
Thanks in advance!
-mico
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
