Methodological clarification

[barneydobson]

Generally is the assumption of the network alignment being conditioned on the street network valid? In my experience in the US, this is in most instances not correct. Has there been any comparison with a real network that can be cited in the manuscript?

Originally posted by cbuahin in #384

[barneydobson]

@cbuahin this is explicitly discussed in the research paper that sets out a lot of the theoretical assumptions behind SWMManywhere. In short though - to paraphrase the paper, we have a few options for deriving potential pipe locations:

• No prior assumptions about where pipes lay (any node may connect to any other node), and instead leave everything down to the network optimization - though this quickly becomes computationally intractable
• Pipes lay under roads precisely (very sensitive to quality of road data)
• Road data can be cleaned to create a more 'pipe-like' layout from the road data - this is what SWMManywhere does.
• Pipes follow hydrological flow paths - if you account for building heights, this ends up being very similar to the road network, if you don't account for building heights you end up with pipes in odd places (though there are some cited references in the paper where others have taken this approach). Further, I argue in the paper that hydrological flow paths are just one of many factors that should be considered in the network optimization, thus it they are more sensibly incorporated as a weighting factor for optimization.

Ultimately the question of where may pipes potentially lay is a very important one, and the current approach is a pragmatic balance of computational feasibility and data availability.

I propose to not address this further for the JOSS review, but I can format this issue to be a discussion point for other proposals on potential pipe-carrying locations.

[cbuahin]

Thanks for sharing the paper. I must admit I missed it.

[barneydobson]

Not your fault - it hadn't been accepted when the JOSS paper was submitted! Though it will be added in #377

[barneydobson]

(discussion can continue in #401 )