Never ending runs on Skylla (sometimes on PC as well) about cmf

[florianjehn]

Still working for beta = 0.3 and 0.25.

[florianjehn]

Beta <= ~ 0.2 seems to trigger the behaviour.

[philippkraft]

Low betas are really difficult for the solver and I can't think of a natural meaning of a beta below 0.5. A do not even have an idea for any beta < 1. The numerical woes of low betas need to be documented, both in the API and in the Kinematic Wave tutorial.

[philippkraft]

From http://fb09-pasig.umwelt.uni-giessen.de/cmf/wiki/CmfTutKinematicWave:
"

A residual water content makes sense eg. for percolation in simplified models where percolation starts
only above field capacity, where the residual water content equals field capacity of the top soil, or it
can be used as the volume behind a dam. To find a justification for β is more challenging: β>1
describes systems, where flow is over linear slow for low storages and gets much faster for a large
storage. Unsaturated flow can be modelled that way, if one takes an analogy with the Brooks-Corey
retention curve, betas from 4 to 15 are reasonable. The standard beta for the "real" kinematic wave
from the Manning-Strickler equation for sheet flow in 53. If you find a real world meaning for β<1, please
inform the authors of cmf. However, for β<0.3 it is reported, that numerical instabilities will arise, even
when using the CVodeIntegrator

"