Why after grdblend ,the grid spacing changed?

Q&A
1

Hello,friends
there has four data is 12.5m spacing,and i use grdblend command to blend them together,and the blendedfile shows it has a spacing of 12.5005167852m.
but the range was right ,it seemed to call a resample command,Why? and how i can get a 12.5m blended file?
Thanks a lot!
the four data ready to be blend:

AP_14636_FBD_F6850_RT1.dem.grd: Pixel node registration used [Cartesian grid]
AP_14636_FBD_F6850_RT1.dem.grd: Grid file format: cs = GMT netCDF format (16-bit integer, deprecated)
AP_14636_FBD_F6850_RT1.dem.grd: x_min: 766676.6875 x_max: 847651.6875 x_inc: 12.5 name:  n_columns: 6478
AP_14636_FBD_F6850_RT1.dem.grd: y_min: 8144894.25 y_max: 8218369.25 y_inc: 12.5 name:  n_rows: 5878
AP_14636_FBD_F6850_RT1.dem.grd: z_min: 36 z_max: 1293 name: meters**
 
AP_14636_FBD_F6860_RT1.dem.grd: Pixel node registration used [Cartesian grid]
AP_14636_FBD_F6860_RT1.dem.grd: Grid file format: cs = GMT netCDF format (16-bit integer, deprecated)
AP_14636_FBD_F6860_RT1.dem.grd: x_min: 754256.6875 x_max: 835469.1875 x_inc: 12.5 name:  n_columns: 6497
AP_14636_FBD_F6860_RT1.dem.grd: y_min: 8200034.25 y_max: 8273359.25 y_inc: 12.5 name:  n_rows: 5866
AP_14636_FBD_F6860_RT1.dem.grd: z_min: 44 z_max: 1292 name: meters
 
AP_15059_FBS_F6850_RT1.dem.grd: Pixel node registration used [Cartesian grid]
AP_15059_FBS_F6850_RT1.dem.grd: Grid file format: cs = GMT netCDF format (16-bit integer, deprecated)
AP_15059_FBS_F6850_RT1.dem.grd: x_min: 824462.9375 x_max: 905437.9375 x_inc: 12.5 name:  n_columns: 6478
AP_15059_FBS_F6850_RT1.dem.grd: y_min: 8144043 y_max: 8217493 y_inc: 12.5 name:  n_rows: 5876
AP_15059_FBS_F6850_RT1.dem.grd: z_min: -193 z_max: 1619 name: meters
 
AP_15059_FBS_F6860_RT1.dem.grd: Pixel node registration used [Cartesian grid]
AP_15059_FBS_F6860_RT1.dem.grd: Grid file format: cs = GMT netCDF format (16-bit integer, deprecated)
AP_15059_FBS_F6860_RT1.dem.grd: x_min: 811712.9375 x_max: 893225.4375 x_inc: 12.5 name:  n_columns: 6521
AP_15059_FBS_F6860_RT1.dem.grd: y_min: 8199403 y_max: 8272753 y_inc: 12.5 name:  n_rows: 5868
AP_15059_FBS_F6860_RT1.dem.grd: z_min: 44 z_max: 1292 name: meters

Blend command:(Notice i used -rp to declare a pixel node registration)

gmt grdblend blend.job -rp -Gblend.grd -R754256.6875/905437.9375/8144043.00/8273359.25

After Blend:

blend.grd: Pixel node registration used [Cartesian grid]
blend.grd: Grid file format: nf = GMT netCDF format (32-bit float), CF-1.7
blend.grd: x_min: 754256.6875 x_max: 905437.9375 x_inc: 12.5005167852 name:  n_columns: 12094
blend.grd: y_min: 8144043 y_max: 8273359.25 y_inc: 12.500362494 name:  n_rows: 10345
blend.grd: z_min: -194.100585938 z_max: 1645.43127441 name: z

The manul told :Any input grid that does not share the final output grid’s node registration and grid spacing will automatically be resampled ,but it obviously not that case.

2

is already non divisible by 12.5 intervals

3

gmt math -Q 905437.9375 754256.6875 SUB gives me 151181.25 which is clearly not a multiple of 12.5.

4

Thanks a lot,Then i tryed another range:
gmt grdblend blend.job -Gblend.grd -R759256.6875/900431.6875<U+202C>/8149043.00/8268355.5<U+202C>
Which is the multiple of 12.5,and the info of result:

blend.grd: Gridline node registration used [Cartesian grid]
blend.grd: Grid file format: nf = GMT netCDF format (32-bit float), CF-1.7
blend.grd: x_min: 759256.6875 x_max: 900437.9375 x_inc: 12.4994466578 name: n_columns: 11296
blend.grd: y_min: 8149043 y_max: 8268359.25 y_inc: 12.5003928759 name: n_rows: 9546
blend.grd: z_min: -164.104415894 z_max: 1642.68981934 name: z
blend.grd: scale_factor: 1 add_offset: 0
blend.grd: format: netCDF-4 chunk_size: 129,129 shuffle: on deflation_level: 3

the height also resampled!

5

Big Thanks! Professor! Then i tryed another range:
gmt grdblend blend.job -Gblend.grd -R759256.6875/900431.6875<U+202C>/8149043.00/8268355.5<U+202C>
Which is the multiple of 12.5,and the info of result:

blend.grd: Gridline node registration used [Cartesian grid]
blend.grd: Grid file format: nf = GMT netCDF format (32-bit float), CF-1.7
blend.grd: x_min: 759256.6875 x_max: 900437.9375 x_inc: 12.4994466578 name: n_columns: 11296
blend.grd: y_min: 8149043 y_max: 8268359.25 y_inc: 12.5003928759 name: n_rows: 9546
blend.grd: z_min: -164.104415894 z_max: 1642.68981934 name: z
blend.grd: scale_factor: 1 add_offset: 0
blend.grd: format: netCDF-4 chunk_size: 129,129 shuffle: on deflation_level: 3

How to choose a right range?
I want the result is 12.5m spacing and non resampled
I tried another software name global mapper and it return the result i wanted,
But i also hope to use gmt to do the same thing!

6

I can confirm the odd result of 12.500xxx. WIll look into it. It is as if floats and not doubles are used.

7

Wait a minute. Your -R is full of hidden control characters that switches the args around. Please the -R again without copy/paste from a web page or word doc or whatever you did. Works fine then.

8

Snipaste_2020-04-24_16-19-27
Snipaste_2020-04-24_16-19-271023×420 5.3 KB

I type them manually,and the result above also not the hoped

9

I think this is the result of a mix between grid and pixel registrations. Sometimes the blend is reported as grid and other as pixel. And about the difference from 12.5, see

(12.5 /2)/11294

ans =

      0.000553391181158137

which is the difference from 12.5 in your case above

10

Thanks Joaquim!
It will be better if there is any resolution.

11

Dear Joaquim:
another question,Please help

Thanks!

12

Not sure what you mean. Would need to check more closely, but it might happen that your case does not have a solution that does not reinterpolate (which is different from the 12.5 increment issue). If grid origins are not at integer steps of 12.5 from each other to do a blend some must be reinterpolated. The non 12.5 issue is a question -R and -r