ddocquie closed issue pbarriat/ecearth_patch#16

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Plots...

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Here are time series of Arctic sea-ice area and volume updated with EC16 in green (run without melt pond and bare puddled ice albedo=0.58, starting in 2060 from the run without melt pond and albedo=0.50 [EC13]). These results show that the effect of reducing the albedo by 8% is probably much weaker than internal variability... FYI, I haven't succeeded in launching EC15 (run with melt pond starting from control without melt pond).

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Here are time series of Arctic sea-ice area and volume updated with EC16 in green (run without melt pond and bare puddled ice albedo=0.58, starting in 2060 from the run without melt pond and albedo=0.50 [EC13]). These results show that the effect of reducing the albedo by 8% is probably much weaker than internal variability... FYI, I haven't succeeded in launching EC15 (run with melt pond starting from control without melt pond).

5 years ago

ddocquie reopened issue pbarriat/ecearth_patch#16

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @pbarriat, I'm trying to launch experiments EC15 and EC16 on zenobe with EC-Earth (/home/acad/ucl-elic/ddocquie/modeles/ecearth_primavera/runtime/classic) starting from year 2060 of an existing experiment EC13 (control run without melt ponds). EC15 is an experiment where I switch on the third melt-pond param. (nn_pnd_scheme=2) and EC16 is an experiment where I switch off melt ponds and albedo of bare puddled ice is increased. EC16 is running without problem, but EC15 crashes after 2-3 seconds of run. It seems that switching on melt ponds (like in EC15) from an experiment where melt ponds are switched off (like in EC13) leads to problems. Here is the end of the log file:

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Issue 643 on the EC-Earth portal: https://dev.ec-earth.org/issues/643?issue_count=184&issue_position=1&next_issue_id=642

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Arctic sea-ice area and volume of these runs and corresponding AMOC.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @sterlin and @pbarriat, FYI my 4 100-yr runs for testing melt ponds with EC-Earth in coupled mode are now all finished (using 2000-yr constant forcing). Each run took about 1 month to run on zenobe with 144 cores (total time including waiting time in the queue). I'll open an issue on the EC-Earth portal related to that and send you the link when done. I'm in the process of transferring the NEMO outputs on pelican: /storepelican/ddocquie/EC-Earth_runs/ECxx, where the runs of interest are EC10, EC11, EC13 and EC14. I haven't saved IFS outputs of these runs. Reminder: EC10 = control without melt ponds and albedo of puddled ice = 0.58; EC11 = run with topographic melt-pond param.; EC13 = control without melt ponds and albedo of puddled ice = 0.50 (original value); EC14 = run with empirical melt-pond param. If you want to have a look at the .sh files used to run those on zenobe: /home/acad/ucl-elic/ddocquie/modeles/ecearth_primavera/runtime/classic.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#20

Yes, that may be the reason. Worth checking with @pbarriat.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#20

Hi @fmasson, this is strange. I've never had problems in restarting from an existing experiment. But if I remember correctly, I always keep the number of processors similar to the 'mother experiment'. You can have a look at my .sh files on zenobe: /home/acad/ucl-elic/ddocquie/modeles/ecearth/runtime/classic. For example, you can compare EC05.sh (150-yr long control run that you're using) and AL00.sh (1-yr long albedo perturbed experiment starting at year 108 of EC05 with reduced albedo, this is one of the members used by Steve in his master thesis). I'll be in LLN tomorrow, we can discuss it.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @sterlin, some updates related to my EC-Earth runs (with fixed 2000-yr forcing). I have now 80 years for my control run without melt-pond param. and albedo=0.58 (EC10), 73 years for my run with the topo param. (MP3, EC11), 48 years for my control run without melt-pond param. and albedo=0.50 (EC13), and 49 years for my run with the empirical param. (MP2, EC14). Attached are March and September Arctic sea-ice areas, as well as April and September Arctic sea-ice volumes. As you can see, the topo param. starts to go below the control after 35 years, but it goes back to the control around 2040 before going below again. As expected, the empirical param. and the control with reduced albedo produce less sea ice. I'll keep you updated once these runs terminate.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @sterlin, some updates related to my EC-Earth runs (with fixed 2000-yr forcing). I have now 80 years for my control run without melt-pond param. and albedo=0.58 (EC10), 73 years for my run with the topo param. (MP3, EC11), 48 years for my control run without melt-pond param. and albedo=0.50 (EC13), and 49 years for my run with the empirical param. (MP2, EC14). Attached are March and September Arctic sea-ice areas, as well as April and September Arctic sea-ice volumes. As you can see, the topo param. starts to go below the control after 35 years, but it goes back to the control around 2040 before going below again. As expected, the empirical param. and the control with reduced albedo produce less sea ice. I'll keep you updated once these runs terminate.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#20

Hi @fmasson, sorry but I've never used a different domain decomposition when starting EC-Earth from a pre-existing restart.

5 years ago

ddocquie closed issue pbarriat/ecearth_patch#16

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @sterlin, I've just checked the routine albedo.F90 in the sources. I confirm that rn_alb_imlt (rn_alb_im in the code) is not used when nn_ice_alb=2 (home-made albedo param. of @lecomte). With this home-made param., the surface albedo is computed as a weighted mean of snow, ponds and bare ice contributions, where the bare ice contribution is computed from the bare frozen ice only. So I think this explains why using a different bare puddled ice albedo with the melt-pond param. doesn't change the results.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @pbarriat and @sterlin, good news: having ifs_cmip_fixyear=2000 fixes the problem, my runs continue after 27 years. I have 4 ongoing runs: EC10 (control with albedo of bare puddled ice rn_alb_imlt=0.58), EC11 (topo param. [nn_pnd_scheme=2] with rn_alb_imlt=0.50), EC12 (topo param. with rn_alb_imlt=0.58), and EC14 (control with rn_alb_imlt=0.50). Below are Arctic sea-ice area and volume time series (28 years for EC10, 24 years for EC11 and 15 years for EC12). As you can see, in the first years of these simulations, the topo param. (EC11 and EC12) produces more sea ice than the control (consistent with SMHI results). I guess that it will start going below the control at some point. Note that EC11 and EC12 produce exactly the same results, meaning (I think) that the albedo of puddled ice is not used when melt-pond params. are switched on. Jean, could you check that please? If yes, I think I can stop EC12.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @pbarriat and @sterlin, FYI tests are ongoing on zenobe with ifs_cmip_fixyear=2000 for both a control run without melt ponds and enhanced albedo of puddled ice (a=0.58) (exp. EC10) and a run with the topographic melt-pond param. and no change in puddled ice albedo (a=0.50) (exp. EC11). Below are time series of SIA and SIV for the first 8 yrs (EC10) and first 4 yrs (EC11). Slight reduction in SIA and SIV in the first yrs of simulation.

5 years ago

ddocquie commented on issue pbarriat/ecearth_patch#16

Hi @pbarriat, I'll let you know if I pass the 27-yr threshold... Maybe the problem was just that I didn't fix the year.

5 years ago