Heads calculation: expose 'neglect_steady' in headalongline

[MattBrst]

In the transient head function there is the keyword neglect_steady=False:

timflow/timflow/transient/model.py

Line 359 in 882bc67

def head(self, x, y, t, layers=None, aq=None, derivative=0, neglect_steady=False):

I would like to expose that keyword as well in the transient headalongline function. Is this OK for you guys?

My use case:

1. I am working on transient 1D models, and want to load a steady model as well.
2. Since the client is interested in responsefactors, I need to have the steady and transient responses both separated and combined.

[dbrakenhoff]

Makes sense to me :).

[mbakker7]

This is an interesting thought. If I understand it correctly, what you propose is that when a steady model is added to a transient model with Model(steady=...), then you still want to be able to compute the headalongline for just the steady solution. If we implement a neglect_steady option, then we have to do that for a bunch of functions (head, contour, trace, etc.). So I am not sure this is the way to go.

What we are thinking off for the future is to have one "combined" model that includes a steady model and a transient model. For that general model it may indeed be useful to be able to get "only" steady or "only" transient heads. Although I see this was already merged into dev and was already present in head. So this is more of a warning, that this use case will likely change in the future.

[MattBrst]

Small correction on Mark's interpretation:

1. I create a steady model;
2. I create a transient model and load the steady model in that model;
3. I can calculate the steady component with my steady model;
4. I want to be able to compute transient results without the steady component included.

My current setup is that I have two separate models (1: steady, 2: transient without steady) and have to add them when I need them combined.

So the reason for asking was exactly to get to know your thoughts about future developments. Maybe for now I'll just stick with my current setup.

[dbrakenhoff]

My bad, I thought it was a small change and it made sense :). The way I interpreted it is that when you compute heads with a transient model with a steady model set, you want either only transient heads (neglect_steady=True) or transient+steady heads (neglect_steady=False). For just steady heads you would have to use the steady model.

But agreed that if we implement this here, it would be consistent to do this across all functions that access heads which is maybe not what we want given the ideas about a future combined model class and a different way to access the submodels.

I'd be fine reverting the change, but I also think it is pretty harmless (and maybe useful for Mattijs at the moment). So I agree with your last thought, that we can accept this for now, but it will probably change gain soon.

[MattBrst]

Actually I agree with Mark's thoughts that when we are having one combined model, it would be nice to be able to choose between steady/transient/both components.

I would be fine reverting the change.