It looks like the first image is using the WSG84 geodetic reference system, whereas the second image is using GRS80, which is why is looks curved. This suggests the objects haven't been converted to the same CRS.
Can you provide the climate data, or a sample, so I or another user can look into this further?
The projection methods commonly used in Australia are Albers equal area, Lambert conformal conic and transverse mercator.
Since the first two need one or two latitudes as parameter, I assume that transverse mercator will fit. UTM uses the same method.
You may take EPSG:3113 GDA94/BCSG02 as an example:
+proj=tmerc +lat_0=-28 +lon_0=153 +k=0.9999900000000001 +x_0=50000 +y_0=100000 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs
and exchange the parameters with those you have got. Make sure to use decimal degrees for the central meridian. lat_0 seems to be missing, so you might try the equator with +lat_0=0
.
Basically, the units could be something else than meters, but this seems to be unusual in Australia.
From the extended information you give, the projection string should be:
+proj=tmerc +lat_0=0 +lon_0=120.95 +k=0.9999 +x_0=446904.02 +y_0=2879827.84 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs
Your Origin point leads me to a point east of the town named Newman, West Australia.
and the full command line:
gdalwarp -s_srs "+proj=tmerc +lat_0=0 +lon_0=120.95 +k=0.9999 +x_0=4 46904.02 +y_0=2879827.84 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs" -t_srs EPSG:28350 im.tif out.tif
Best Answer
I got it working in two steps for the first band:
This does not use the geoloc information in subdataset 2 and 3, but the corner coordinates from those subdatasets, converted to aeqd with
and added 500m offset from mid cell to the edge. I omitted
datum=WGS84
from the projection because this conflicts with the given radius.This is my result, with the corner points from the subdatasets 2 and 3:
For all 5545 bands, you better use a vrt file:
but the resulting tif is 2.4GB even with compression.