Recurrent MARBL WARNING in BHIST and BHISTcmip6 cases

[acgvar]

Hello!
While tailing the cesm.log files while running the compsets BHIST and BHISTcmip6 with f19 resolution, there is a recurrent warning which states similar to this,

(Task 476, block 1) MARBL WARNING (marbl_interior_tendency_mod:compute_large_detritus_prod): dz*DOP_loss_P_bal= 0.135E-012 exceeds Jint_Ptot_thres= 0.271E-013

What is causing the above? Is there a quick remedy? Is dt one possible solution?

 

[Yao]

Hi, Alvin. Have you shoveled this problem? I meet the same problem.

Hello!
While tailing the cesm.log files while running the compsets BHIST and BHISTcmip6 with f19 resolution, there is a recurrent warning which states similar to this,

(Task 476, block 1) MARBL WARNING (marbl_interior_tendency_mod:compute_large_detritus_prod): dz*DOP_loss_P_bal= 0.135E-012 exceeds Jint_Ptot_thres= 0.271E-013

What is causing the above? Is there a quick remedy? Is dt one possible solution?

 

[acgvar]

Hi Yao,

Not really. The simulation continued with just the warning. The simulation completed anyway.
I still wonder.

 

[jeffhu]

Same problem

 

[klindsay]

Autotrophic plankton in MARBL have variable phosphorus:carbon (P:C) ratios in their biomass, while grazers have a fixed P:C ratio. If the grazers consume plankton whose P:C ratio is less than the grazer P:C ratio, P is removed from the sinking organic P pool (POP) to meet the P requirements of the grazer, i.e., to maintain the grazer's higher P:C ratio. If there is not sufficient P in the POP pool to meet this demand, then P is removed from the dissolved organic P pool (DOP) to meet the P requirements. An early version of the model didn't have this latter process, and total P was not being conserved. When we adding routing from DOP to meet this need, we added a warning message that flags when omitting this routing would have led to a conservation issue. These messages are informational only. There are admittedly too many of them, and we should adjust the code to reduce their occurrence rate. I've just filed an issue ticket in the github repo for MARBL to look into this.