Setup missing connections based on flow direction

[xldeltares]

• [x] #61
• [ ] #62

[xldeltares]

[xldeltares]

Discussion with Helene: There is already function that converts flow direction into geodataframe. check the below documentation. Then the convertion into graphs can an easy next step. https://deltares.github.io/hydromt_wflow/latest/_examples/prepare_ldd.html

[xldeltares]

Discussion with Helene: During this step, add flag: lines that are added (approximate), lines that are from the vector layer. knowledge about percentage of reliability, give indication to the user how much work still needs to be done. e.g. 50% guessed, that means a lot of follow-up dataset work is needed (useful for proposal estimation) the artificially connected line, we can condition on the length which gives estimation of uncertainty.

[xldeltares]

1st draft (updated on 18--8-2023) A function that is called 'setup_graph_from_hydrography` is drafted. This function reads the hydrography data and get flow direction as a graph.

TODO

• [ ] move this issue to later. From experiences working with flow direction, deriving flwdir from dem is not that straight forward and therefore, wflow requires the user to prepare flwdir under supervision themselves. Because it is cell to cell direction, it makes sense to use this info later for smaller graph arborescence. While for larger graph elements, the direction of the lines are easier to obtain. (suggestions from Helene)
• [ ] continuing previous point, think about upscaling raster to test. Compare that with what real was there for bangkok
• [ ] good to have a sloped test case as well.
• [ ] later: combine the two graphs (from osm and from hydrography) to get directions of the osm graph.