You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Add more system tests covering water tracers/isotopes #495
Very soon, I plan to add a one-timestep debug system test covering the tracer consistency check. Later, we should add more tests, covering at least the following (note that some of these things can be combined so that we only need a couple of tests total):
Longer than one timestep
Exact restart / processor / thread counts: ERP test
Non-cold-start
Multiple compilers
The real isotope configuration (rather than just adding the tracers used for the tracer consistency check)
A test that exercises dynamic landunit code, either by having a multi-year test or by having a multi-day test that turns on the for-testing daily glacier dynamics
A test that includes FATES
I anticipate this being an issue for total_plant_stored_h2o_col: this variable is only non-zero when using FATES, and we'll need to do something to ensure that the tracer version of this is set appropriately. There may be other biogeophysics updates in FATES for which we need to handle tracers.
If this is non-trivial to get working, we could open a separate issue to deal with this later.
Also want to cover some physics options that aren't covered by default, but which impact water fluxes / state updates - for example, groundwater irrigation.
I'm adding a LWISO test to ensure that turning on water isotopes doesn't change answers. However, I'm currently running into problems trying to run multiple time steps with enable_water_isotopes=.true.. I haven't dug into the problem in detail, but I have a feeling it's because we haven't fully tracerized the code yet, and while the test tracers are reset each time step, the true water tracers are not. So for now I'm using enable_water_tracer_consistency_checks=.true..
Once the true water isotopes are fully working, this LWISO test should probably be changed to use true water isotopes rather than test tracers, since that's a more appropriate / useful test (though it probably doesn't matter hugely).
Also, for now we need to use coldStart for this because we can't use initial conditions from a non-wiso case in a case with water isotopes; when we can eventually do that, we should change this test to non-coldStart and we can shorten it somewhat from its current 10 days (I'm making it 10 days in order to trigger more conditions than a 1 day test does when starting from cold start). Perhaps we should make it a 5-day, transient decStart test at that point.
Activity
Add a system test that exercises the tracer consistency check
billsacks commentedon Dec 11, 2018
Also want to cover some physics options that aren't covered by default, but which impact water fluxes / state updates - for example, groundwater irrigation.
Add a system test that exercises the tracer consistency check
billsacks commentedon Jul 15, 2019
I'm changing the one-timestep test to one day, now that I have a mechanism in place to allow bypassing not-yet-tracerized code.
billsacks commentedon Jul 30, 2019
I'm adding a LWISO test to ensure that turning on water isotopes doesn't change answers. However, I'm currently running into problems trying to run multiple time steps with
enable_water_isotopes=.true.. I haven't dug into the problem in detail, but I have a feeling it's because we haven't fully tracerized the code yet, and while the test tracers are reset each time step, the true water tracers are not. So for now I'm usingenable_water_tracer_consistency_checks=.true..Once the true water isotopes are fully working, this LWISO test should probably be changed to use true water isotopes rather than test tracers, since that's a more appropriate / useful test (though it probably doesn't matter hugely).
Also, for now we need to use coldStart for this because we can't use initial conditions from a non-wiso case in a case with water isotopes; when we can eventually do that, we should change this test to non-coldStart and we can shorten it somewhat from its current 10 days (I'm making it 10 days in order to trigger more conditions than a 1 day test does when starting from cold start). Perhaps we should make it a 5-day, transient decStart test at that point.
billsacks commentedon Aug 22, 2019
If we want to add nag tests, see #763 and #789 for caveats.