each dimension and 1D variable must have the same name: not documented, no warning

[Thomas-Ulrich]

This is the file I try to parse:

!Switch
[d_c]: !AffineMap
    matrix:
      x: [1.0, 0.0, 0.0]
      y: [0.0, 0.0, -1.0]
    translation:
      x: 20e3
      y: 0.0
    components: 
      -  !ASAGI
          file: test.nc
          parameters: [d_c]
          var: h

And this is how I create the example netcdf

import numpy as np
dx=5e3
x=np.arange(0,40e3+dx, dx)
y=np.arange(0,20e3+dx, dx)
nx=x.shape[0]
ny=y.shape[0]
from netCDF4 import Dataset
rootgrp = Dataset("test.nc", "w", format="NETCDF4")
rootgrp.createDimension("a", nx)
rootgrp.createDimension("b", ny)
vx = rootgrp.createVariable("a","f4",("a",))
vx[:]=x
vy = rootgrp.createVariable("b","f4",("b",))
vy[:]=y
vplastCo =  rootgrp.createVariable("h","f4",("b","a"))
vplastCo[:,:] = np.ones((ny, nx))
rootgrp.close()

structure of the netcdf

ncdump -h test.nc 
netcdf test {
dimensions:
	a = 9 ;
	b = 5 ;
variables:
	float a(a) ;
	float b(b) ;
	float h(b, a) ;
}

I try then to parse the yaml file use a cpp program, and I get the expected value:

../EvaluateVpVs/EvaluateDc test2.yaml -15000. 0 -5000.
1

On the other hand, when using a netcdf file with this structure:

ncdump -h test.nc 
netcdf test {
dimensions:
	na = 9 ;
	nb = 5 ;
variables:
	float a(na) ;
	float b(nb) ;
	float h(nb, na) ;
}

I get the following error:

../EvaluateVpVs/EvaluateDc test2.yaml -15000. 0 -5000.
terminate called after throwing an instance of 'std::runtime_error'
  what():  test2.yaml@9: Could not find model for point [ 5000 5000 ] in group 1.
Aborted (core dumped)

Which means that asagi only works when each dimension and 1D variable have the same name.
This should be Documented, or a warning should be raised when not the case.

[rettenbs]

This is part of the COARDS Convention for netCDF files: https://ferret.pmel.noaa.gov/Ferret/documentation/coards-netcdf-conventions

Coordinate Variables:
1-dimensional netCDF variables whose dimension names are identical to their variable names are regarded as "coordinate variables" (axes of the underlying grid structure of other variables defined on this dimension).

Probably everything is working fine but the your arrays a and b are no longer considered as coordinate variables and the point 5000, 5000 is out of bounds.

[Thomas-Ulrich]

Hi Sebastian, I see! I missed this important point of the COARDS convention. Then we will make that clear in easi or seissol documentation. Thanks for the feedback! Best, Thomas. Le 19 octobre 2020 19:07:45 GMT+02:00, Sebastian Rettenberger <notifications@github.com> a écrit :

…

This is part of the COARDS Convention for netCDF files: https://ferret.pmel.noaa.gov/Ferret/documentation/coards-netcdf-conventions > **Coordinate Variables:** > 1-dimensional netCDF variables whose dimension names are identical to their variable names are regarded as "coordinate variables" (axes of the underlying grid structure of other variables defined on this dimension). Probably everything is working fine but the your arrays `a` and `b` are no longer considered as coordinate variables and the point 5000, 5000 is out of bounds. -- You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub: #10 (comment)