Lack of consistency between fwd and fwd_intermediate azimuth output

[stephenwinn16]

Problem description

I've noticed an inconsistency between the azimuth output of fwd and fwd_intermediate. In the example below, I compute the lon/lat/azi when moving 1km from London with an initial forward azimuth of 45 degrees with the both methods. As expected from the documentation, fwd returns the backward azimuth whereas fwd_intermediate returns the forward azimuth of the terminus point. For consistency and to avoid confusion, I suggest fwd_intermediate also return the backward azimuth.

Code Sample

from pyproj import Geod, enums

wgs84 = Geod(ellps="WGS84")

# Inputs 
london = (-0.118092, 51.509865) #lon/lat
dist = 1e3  # m
bearing = 45  # deg
flags = enums.GeodIntermediateFlag(0).AZIS_KEEP  # get fwd_intermediate to return azimuths

# fwd
lon, lat, azi = wgs84.fwd(*london, bearing, dist)

# fwd_intermediate (return first and last point)
res = wgs84.fwd_intermediate(*london, bearing, 2, dist, flags=flags, terminus_idx=0, initial_idx=0)

print("lon, lat, azimuth from fwd             :", lon, lat, azi)
print("lon, lat, azimuth from fwd_intermediate:", res.lons[-1], res.lats[-1], res.azis[-1])

which outputs:

lon, lat, azimuth from fwd             : -0.10790547005142687 51.51622011204643 -134.99202649526126
lon, lat, azimuth from fwd_intermediate: -0.10790547005142687 51.51622011204643 45.007973504738736

Environment Information

pyproj info:
pyproj: 3.4.0
PROJ: 9.1.0

Installation method

Installed via pip

[snowman2]

cc: @idanmiara

[snowman2]

Related #841

[idanmiara]

Hi,
As we don't want to break backwards compatibility, I've added this option as a non-default argument.
Hope it helps :)

[stephenwinn16]

Hi @idanmiara,
Thanks for the changes. I understand the argument for backward compatibility. However, wouldn't it make more sense for the default behaviour of 'fwd' and 'fwd_intermediate' to be the same? You could have 'return_back_azimuth' is True by default and a note in the next release warning prior user.

[idanmiara]

Hi @stephenwinn16,
I really don't mind to go either way or have any strong opinion.
I guess it should be @snowman2 's call.

[idanmiara]

Hi @stephenwinn16, @snowman2
I've added return_back_azimuth to fwd() as well.
I would have made the default to return forward azimuth for both, but it doesn't make sense to change the default behavior of fwd so I'm not sure if we would prefer to preserve backwards compatibility with fwd_intermediate or to make the default value of return_back_azimuth=True for the sake of compatibility with fwd.
WDYT ?

[snowman2]

I see this comment in the code:

           # back azimuth needs to be flipped 180 degrees
           # to match what PROJ geod utility produces.

This makes me think that the default return should be the back azimuth for consistency with PROJ. Thoughts?

[idanmiara]

I see this comment in the code:

           # back azimuth needs to be flipped 180 degrees
           # to match what PROJ geod utility produces.

This makes me think that the default return should be the back azimuth for consistency with PROJ. Thoughts?

makes sense, but I guess that the comment should be fixed:
by default, forward azimuth needs to be flipped 180 degrees to match the (back azimuth) output of PROJ geod utilities.

[idanmiara]

Hi @idanmiara, Thanks for the changes. I understand the argument for backward compatibility. However, wouldn't it make more sense for the default behaviour of 'fwd' and 'fwd_intermediate' to be the same? You could have 'return_back_azimuth' is True by default and a note in the next release warning prior user.

@snowman2 agrees with you, so I've changed the default.
I added this parameter to fwd, inv, fwd_intermediate, inv_intermediate

[stephenwinn16]

This makes me think that the default return should be the back azimuth for consistency with PROJ. Thoughts?

Agreed.

@snowman2 agrees with you, so I've changed the default.

Great, this should make everything consistent. Thank you both!