Hacker News Reader

This means everywhere uphill from this creek can't be assigned to a watershed, or we can't model watersheds as a tree. My guess is water modeling software can't handle this situation.

(Cases where an elegant model covers almost everything but not quite are frustratingly common.)

View on HN

I think only if watersheds are required to terminate on an ocean, but that does not work for basins that are not connected to the sea, one notable example of which is not that far south of this point, containing the Great Salt lake. An alternative view (touched on in the article) is that technically, the fork is the end-point of the Atlantic-Pacific divide, and everything upstream on this creek is in a watershed that has one end on one branch downstream of the fork, and the other end on the other branch. In this view, it is the paths downstream of the fork that are not in a watershed. This is not a useful viewpoint, so I agree that this is a corner case in which the tree model does not work.

The Casiquiare canal (it's natural, despite its name) connects the Orinoco and Amazon river systems during peak waterflow system. That's #4 and #1 in terms of discharge, and they don't form a clean water divide.

Reply on news.ycombinator.com