"Impossible case2 in instratus_condensate" in CAM5 under high CO2

[jiangzhu]

Hello,My simulations using CESM1.2.2.1 CAM5 crashed under high CO2 concentration (~24x preindustrial CO2, 6832.8 ppmv) with the following error message in cesm.log: 460: Impossible case2 in instratus_condensate
 460:   1.00000000000000       0.000000000000000E+000  0.000000000000000E+000
 460:                     NaN                     NaN  This problem seems persistent and shows up as long as the global mean temperature reaches ~45 degC. So far, I have reproduced this model failure with 24xCO2 slab ocean (E compset) simulation under the present-day boundary conditions, and with 12xCO2 slab ocean (E compset) and fully coupled (B compset) simulations under the Eocene boundary conditions. Also, I tried to change the model timestep (ATM_NCPL: 48--> 96). The model did run further to 55°C but died again with the same error.Note that the deep time bug fixes have already been patched to the code base (I am actually using pcesm0.2_cesm1_2_2_1).Please help me look into this problem. Thank you very much! Regards,Jiang

 

[hannay]

This is a tricky issue. With very large values of CO2, you are putting the model in a configuration that it hasn't been tested for. Unfortunately, there is not a quick answer for this. 

 

[hannay]

This is a tricky issue. With very large values of CO2, you are putting the model in a configuration that it hasn't been tested for. Unfortunately, there is not a quick answer for this. 

 

[hannay]

This is a tricky issue. With very large values of CO2, you are putting the model in a configuration that it hasn't been tested for. Unfortunately, there is not a quick answer for this. 

 

[jiangzhu]

Dear Cécile,Thank you so much for your reply. I am adding more information here. Hopefully, someone can kindly share some insights on what to look into next.1. Low cloud decreases greatly to only ~20% in the high CO2 experiment (12x PI value).2. Water is no longer conserved. CAM5 is destroying water at an average rate of 0.003 mm/day in the end of the simulation.3. With the same amount of CO2 (12x PI value) and boundary conditions, CAM4 runs OK with a stable surface temperature around 33 degC.  I know some groups have added huge amount of CO2 (~1 bar) to CAM4 and produced stable global temperature close to the boiling point. So, I think this problem was introduced since CAM5.4. Over the deep tropics, the large scale precipitation (maximum of 42 mm/day) is much larger than convective precipitation (maximum of 17 mm/day).Please let me know, if anyone wants to look at more diagnostics. Any help would be greatly appreciated!

 

[jiangzhu]

Dear Cécile,Thank you so much for your reply. I am adding more information here. Hopefully, someone can kindly share some insights on what to look into next.1. Low cloud decreases greatly to only ~20% in the high CO2 experiment (12x PI value).2. Water is no longer conserved. CAM5 is destroying water at an average rate of 0.003 mm/day in the end of the simulation.3. With the same amount of CO2 (12x PI value) and boundary conditions, CAM4 runs OK with a stable surface temperature around 33 degC.  I know some groups have added huge amount of CO2 (~1 bar) to CAM4 and produced stable global temperature close to the boiling point. So, I think this problem was introduced since CAM5.4. Over the deep tropics, the large scale precipitation (maximum of 42 mm/day) is much larger than convective precipitation (maximum of 17 mm/day).Please let me know, if anyone wants to look at more diagnostics. Any help would be greatly appreciated!

 

[jiangzhu]

Dear Cécile,Thank you so much for your reply. I am adding more information here. Hopefully, someone can kindly share some insights on what to look into next.1. Low cloud decreases greatly to only ~20% in the high CO2 experiment (12x PI value).2. Water is no longer conserved. CAM5 is destroying water at an average rate of 0.003 mm/day in the end of the simulation.3. With the same amount of CO2 (12x PI value) and boundary conditions, CAM4 runs OK with a stable surface temperature around 33 degC.  I know some groups have added huge amount of CO2 (~1 bar) to CAM4 and produced stable global temperature close to the boiling point. So, I think this problem was introduced since CAM5.4. Over the deep tropics, the large scale precipitation (maximum of 42 mm/day) is much larger than convective precipitation (maximum of 17 mm/day).Please let me know, if anyone wants to look at more diagnostics. Any help would be greatly appreciated!

 

[jiangzhu]

Dear Cécile,Thank you so much for your reply. I am adding more information here. Hopefully, someone can kindly share some insights on what to look into next.1. Low cloud decreases greatly to only ~20% in the high CO2 experiment (12x PI value).2. Water is no longer conserved. CAM5 is destroying water at an average rate of 0.003 mm/day in the end of the simulation.3. With the same amount of CO2 (12x PI value) and boundary conditions, CAM4 runs OK with a stable surface temperature around 33 degC.  I know some groups have added huge amount of CO2 (~1 bar) to CAM4 and produced stable global temperature close to the boiling point. So, I think this problem was introduced since CAM5.4. Over the deep tropics, the large scale precipitation (maximum of 42 mm/day) is much larger than convective precipitation (maximum of 17 mm/day).Please let me know, if anyone wants to look at more diagnostics. Any help would be greatly appreciated!

 

[rneale]

If I were looking for something, I would drill down into tendencies from individual processes to see how the budgets of water (v,l,i) and temperatrure/energy are changing with changed CO2 content. Do they actually make sense, and are previously small tendency terms now dominating? These kind of problems inevitably are difficult to track down and require a lot of work.Also it is worth noting that becasue CAM4 was able to go to 1 bar it does not mean that CAM4 is right and CAM5 is wrong. It may most likely mean that processes are missing in CAM4 and those simulations should actually be failing at close to boiling point. It goes without saying that the model was never tested in these extreme environments, so model usage is always buyer beware. 

 

[rneale]

If I were looking for something, I would drill down into tendencies from individual processes to see how the budgets of water (v,l,i) and temperatrure/energy are changing with changed CO2 content. Do they actually make sense, and are previously small tendency terms now dominating? These kind of problems inevitably are difficult to track down and require a lot of work.Also it is worth noting that becasue CAM4 was able to go to 1 bar it does not mean that CAM4 is right and CAM5 is wrong. It may most likely mean that processes are missing in CAM4 and those simulations should actually be failing at close to boiling point. It goes without saying that the model was never tested in these extreme environments, so model usage is always buyer beware. 

 

[rneale]

If I were looking for something, I would drill down into tendencies from individual processes to see how the budgets of water (v,l,i) and temperatrure/energy are changing with changed CO2 content. Do they actually make sense, and are previously small tendency terms now dominating? These kind of problems inevitably are difficult to track down and require a lot of work.Also it is worth noting that becasue CAM4 was able to go to 1 bar it does not mean that CAM4 is right and CAM5 is wrong. It may most likely mean that processes are missing in CAM4 and those simulations should actually be failing at close to boiling point. It goes without saying that the model was never tested in these extreme environments, so model usage is always buyer beware. 

 

[rneale]

If I were looking for something, I would drill down into tendencies from individual processes to see how the budgets of water (v,l,i) and temperatrure/energy are changing with changed CO2 content. Do they actually make sense, and are previously small tendency terms now dominating? These kind of problems inevitably are difficult to track down and require a lot of work.Also it is worth noting that becasue CAM4 was able to go to 1 bar it does not mean that CAM4 is right and CAM5 is wrong. It may most likely mean that processes are missing in CAM4 and those simulations should actually be failing at close to boiling point. It goes without saying that the model was never tested in these extreme environments, so model usage is always buyer beware. 

 

[rneale]

ALso if the precip rate is 42 mm/day from large-scale rain then the Hadley circulation and ITCZ have to be incredibly intense, is this the case?

 

[rneale]

ALso if the precip rate is 42 mm/day from large-scale rain then the Hadley circulation and ITCZ have to be incredibly intense, is this the case?

 

[rneale]

ALso if the precip rate is 42 mm/day from large-scale rain then the Hadley circulation and ITCZ have to be incredibly intense, is this the case?

 

[rneale]

ALso if the precip rate is 42 mm/day from large-scale rain then the Hadley circulation and ITCZ have to be incredibly intense, is this the case?