Geoid undulation always calculated to be zero for version 3.7.2

[ElinaKoester]

Dear developers,

I use pyproj to calculate the geoid undulation at some coordinates where I measured some data. Today I noticed that the undulation is always calculated to be zero when using version 3.7.2 while it looks as expected for version 3.71. varying between -40 and 60 meters over Europe.

Attached you'll find a minimal working example and pictures.

Please excuse If I provide too much or too little information. This is my first bug report ever...

Thank you!

Code Example

[geoid_undulation_bug_report_MWE.py](https://github.com/user-attachments/files/27515221/geoid_undulation_bug_report_MWE.py)

import numpy as np
import matplotlib.pyplot as plt
from pyproj import CRS, Transformer
from mpl_toolkits.basemap import Basemap


def calcOrthometricHeights(lat, lon, ellipsoidal_height):
    '''
    Calculates the orthometric height from a set of GPS coordinates made of lat, lon and ellipsoidal WGS84 height

    Returns the orthometric height (height above mean sea level) using EGM2008
    '''

    wgs84 = CRS("EPSG:4979")
    egm08 = CRS("EPSG:3855")

    # create a transformer from WGS84 to EGM08

    transformer = Transformer.from_crs(wgs84, egm08, always_xy=True)

    # Compute orthometric heights
    orthometric_heights = []
    for lon, lat, h_ell in zip(lon, lat, ellipsoidal_height):
    # Transform (lon, lat, ellipsoidal_height) → (lon, lat, orthometric_height)
        _, _, h_ortho = transformer.transform(lon, lat, h_ell)
        orthometric_heights.append(h_ortho)

    orthometric_height = np.array(orthometric_heights)

    geoid_undulation = ellipsoidal_height -   orthometric_height

    return orthometric_height, geoid_undulation



# Calculate the geoid undulation 

max_lat = 65
min_lat = 20
max_lon = 60
min_lon = -30

lower_left_corner  = (min_lat, min_lon) # lat, lon
upper_right_corner = (max_lat, max_lon)

grid_lat = np.arange(lower_left_corner[0], upper_right_corner[0], 0.1)
grid_lon = np.arange(lower_left_corner[1], upper_right_corner[1], 0.1)

lats, lons = np.meshgrid(grid_lat, grid_lon)


# Calculate the orthometric height at ellipsoidal height 0

ellipsodial_heigth = np.zeros((lats.shape[0], lons.shape[1]))

orthometric_height, undulation = calcOrthometricHeights(lats, lons, ellipsoidal_height=ellipsodial_heigth)

print(undulation)

# Plot the results

lat_min, lat_max = lower_left_corner[0], upper_right_corner[0]
lon_min, lon_max = lower_left_corner[1], upper_right_corner[1]

fig, axs = plt.subplots(1, 1, figsize=(15, 10), constrained_layout=True)

ax = axs
m = Basemap(projection='cyl',
            llcrnrlat=lat_min, urcrnrlat=lat_max,
            llcrnrlon=lon_min, urcrnrlon=lon_max,
            resolution='i', ax=ax)
m.drawcoastlines()
m.drawcountries()
m.drawparallels(np.arange(lat_min, lat_max+1, 5), labels=[1,0,0,0], fontsize=8)
m.drawmeridians(np.arange(lon_min, lon_max+1, 10), labels=[0,0,0,1], fontsize=8)

cs = m.pcolormesh(lons, lats, undulation, shading='auto', cmap='turbo', latlon=True)
ax.set_title('Geoid undulation (EGM2008)', fontsize=12)
cbar = fig.colorbar(cs, ax=ax, orientation='vertical', pad=0.02, aspect=30, shrink=0.8)
cbar.set_label('Geoid undulation (m)', fontsize=10)
cbar.ax.tick_params(labelsize=8)

plt.show()

Problem description

For version 3.7.1 the undulation looks as expected, varying between -40 and 60 meters in Europe:

For version 3.7.2 the undulation is always 0, which is wrong.

Expected Output

The undulation should look like for version 3.7.1

Installation method

Miniforge

Conda environment information (if you installed with conda):

$ conda list pyproj

# Name                     Version          Build            Channel
pyproj                     3.7.2            py313hbf73894_3  conda-forge

[jjimenezshaw]

What happens if you try egm08 = CRS("EPSG:4326+3855")? Or EPSG:9518?

IIRC, there was a change in the database regarding the usage of the file Und_min1x1_egm2008_isw=82_WGS84_TideFree ,that is not included in cdn.proj.org

[ElinaKoester]

Thank you for your fast answer!

I noticed the issue with using different computers. One had the old version installed, which did not produce the error. On the other computer the new version is installed, which produces the error.

Downgrading the version on the second computer did not help, installing through pip also did not change it. But I saw that the miniforge version on the computer which produces the error is newer.
On the computer which does not produce the error, the miniforge version is older. Upgrading to 3.7.2 still calculates the right undulation.

In short: It seems to me that the version is not the issue, but in which miniforge version it is installed. Could this be a thing?

Using egm08 = CRS("EPSG:4326+3855") does not change anything when using the 'errorus' version.

Here is the minforge information for both computers:

Results in error (new miniforge)

(windretrieval2) C:\>conda info

     active environment : windretrieval2
    active env location : C:\Users\koes_el\AppData\Local\miniforge3\envs\windretrieval2
            shell level : 2
       user config file : C:\Users\koes_el\.condarc
 populated config files : C:\Users\koes_el\AppData\Local\miniforge3\.condarc
          conda version : 26.3.2
    conda-build version : not installed
         python version : 3.12.12.final.0
                 solver : libmamba (default)
       virtual packages : __archspec=1=skylake
                          __conda=26.3.2=0
                          __win=10.0.22631=0
       base environment : C:\Users\koes_el\AppData\Local\miniforge3  (writable)
      conda av data dir : C:\Users\koes_el\AppData\Local\miniforge3\etc\conda
  conda av metadata url : None
           channel URLs : https://conda.anaconda.org/conda-forge/win-64
                          https://conda.anaconda.org/conda-forge/noarch
          package cache : C:\Users\koes_el\AppData\Local\miniforge3\pkgs
                          C:\Users\koes_el\.conda\pkgs
                          C:\Users\koes_el\AppData\Local\conda\conda\pkgs
       envs directories : C:\Users\koes_el\AppData\Local\miniforge3\envs
                          C:\Users\koes_el\.conda\envs
                          C:\Users\koes_el\AppData\Local\conda\conda\envs
    temporary directory : C:\Users\koes_el\AppData\Local\Temp
               platform : win-64
             user-agent : conda/26.3.2 requests/2.32.5 CPython/3.12.12 Windows/11 Windows/10.0.22631 solver/libmamba conda-libmamba-solver/25.11.0 libmambapy/2.4.0
          administrator : False
             netrc file : None
           offline mode : False

Does not result in error (old miniforge version)

(windretrieval) C:\>conda info

     active environment : windretrieval
    active env location : C:\Users\koes_el\AppData\Local\miniforge3\envs\windretrieval
            shell level : 2
       user config file : C:\Users\koes_el\.condarc
 populated config files : C:\Users\koes_el\AppData\Local\miniforge3\.condarc
          conda version : 24.11.3
    conda-build version : not installed
         python version : 3.12.5.final.0
                 solver : libmamba (default)
       virtual packages : __archspec=1=skylake
                          __conda=24.11.3=0
                          __win=0=0
       base environment : C:\Users\koes_el\AppData\Local\miniforge3  (writable)
      conda av data dir : C:\Users\koes_el\AppData\Local\miniforge3\etc\conda
  conda av metadata url : None
           channel URLs : https://conda.anaconda.org/conda-forge/win-64
                          https://conda.anaconda.org/conda-forge/noarch
          package cache : C:\Users\koes_el\AppData\Local\miniforge3\pkgs
                          C:\Users\koes_el\.conda\pkgs
                          C:\Users\koes_el\AppData\Local\conda\conda\pkgs
       envs directories : C:\Users\koes_el\AppData\Local\miniforge3\envs
                          C:\Users\koes_el\.conda\envs
                          C:\Users\koes_el\AppData\Local\conda\conda\envs
               platform : win-64
             user-agent : conda/24.11.3 requests/2.32.3 CPython/3.12.5 Windows/11 Windows/10.0.22631 solver/libmamba conda-libmamba-solver/24.7.0 libmambapy/1.5.9
          administrator : False
             netrc file : None
           offline mode : False

Thanks for your help!

[jjimenezshaw]

The difference is probably in PROJ. What are the PROJ versions used?

[ElinaKoester]

Weirdly enough, the proj versions seems to be the same on both computers:

Wrong undulation (new miniforge)

(windretrieval2) C:\>conda list proj
# packages in environment at C:\Users\koes_el\AppData\Local\miniforge3\envs\windretrieval2:
#
# Name                     Version          Build            Channel
proj                       9.7.1            hd30e2cd_3       conda-forge
pyproj                     3.7.2            py313hbf73894_3  conda-forge

Right undulation (old miniforge)

(windretrieval) C:\>conda list proj
# packages in environment at C:\Users\koes_el\AppData\Local\miniforge3\envs\windretrieval:
#
# Name                    Version                   Build  Channel
proj                      9.7.1                hd30e2cd_3    conda-forge
pyproj                    3.7.2           py313hbf73894_3    conda-forge

I have updated all packages on both environments to the newest versions. The only difference I see is the miniforge version, but I do not understand how this would affect the packages? It should not affect the packages, right?

Thank you for your help.

[snowman2]

In conda, PROJ_NETWORK is always enabled by default. Using pip, it is not. That is likely the difference you are encountering.

[ElinaKoester]

Right now everything is installed via conda (miniforge) on both machines. Should I install pyproj and proj via pip instead? Only installing only pyproj via pip did not change the error.

[ElinaKoester]

Now I installed proj and pyproj via pip on both machines and on the computer with the old minforge the calculation is still correct, while on the computer with the new miniforge the undulation is still zero. What else could I try? I am really at my wits end right now and wonder if I overlook something.

[ElinaKoester]

I realized that when I call the transformer on the computer where the code only returns zero

wgs84 = CRS("EPSG:4979")
egm08 = CRS("EPSG:3855")

# create a transformer from WGS84 to EGM08

transformer = Transformer.from_crs(wgs84, egm08, always_xy=True)

I get the following information:

<Concatenated Operation Transformer: axisswap>
Description: axis order change (2D) + Inverse of Transformation from EGM2008 height to WGS 84 (ballpark vertical transformation, without ellipsoid height to vertical height correction)
Area of Use:
- name: World.
- bounds: (-180.0, -90.0, 180.0, 90.0)

On the other computer, where the geoid undulation is not zero, the (ballpark vertical transformation, without ellipsoid height to vertical height correction) statement is not stated. And it indeed returns the vertical height/ geoid undulation.

Now I wonder if I might be lacking some file? Or do I have to enable settings on the first computer to be able to calculate the vertical height?

Note that on both computers pyproj and proj are now installed with pip. Thank you!