search

UW-Hydro / VIC

The Variable Infiltration Capacity (VIC) Macroscale Hydrologic Model
http://vic.readthedocs.io
MIT License
261 stars 380 forks source link

Disallow explicit finite difference soil temperature scheme #37

Open tbohn opened 10 years ago

tbohn commented 10 years ago

The explicit scheme (currently the default) is not accurate for certain combinations of time step and node spacing. Currently VIC gives an error message if the user attempts to run the explicit scheme with these combinations. Given that the implicit scheme (currently optional) has no such restrictions and does not require substantially more cpu time than the explicit scheme, it would seem beneficial to remove the explicit scheme.

tbohn commented 10 years ago

Another advantage of removing the explicit scheme would be a major simplification of the soil temperature code.

tbohn commented 10 years ago

Issue #66 would make the implicit scheme the default, without getting rid of the explicit scheme entirely. I'm suggesting issue #66 for VIC 4.2, with the idea that removing the explicit scheme entirely could happen at any later time.

I had forgotten about this issue when I opened issue #66, but the two issues potentially could be slated for different releases. Sorry if this is causing needless issue churn. Please feel free to rearrange/combine any of these issues as you see fit.

tbohn commented 8 years ago

I don't understand - you really want to remove the explicit scheme for 5.0? We still don't have a fix for frozen soil behavior; Tara Troy's code only works in the explicit mode. I really don't think we can do this for 5.0.

bartnijssen commented 8 years ago

@tbohn: Joe and are sitting/standing here laughing given your comment "I don't understand ..." in an issue that only has previous comments from you. Imagine how the rest of us feel :-)

tbohn commented 8 years ago

Sorry about any confusion. What exactly do you want to have happen? Just insert logic that exits with error if explicit is selected (easy) or remove all explicit-mode code from the model (harder)? I'm worried that this is a potential can of worms that we don't want to open before the release of 5.0.

jhamman commented 8 years ago

Agreed. We can move this to 5.1. This isn't related to the refactor and is best placed in 5.1.