Meeting Reports

On Wednesday, 22th November

With everyone.

Comparison of lifetimes between run of Prather with and without nudging. Try to understand results for the coupled model (fully coupled, only ocean/atmosphere coupling, only land/atmosphere coupling, no coupling). It stresses on the initial state (restarts).

TO DO

• Interpolate the oxidants in the CMIP6 simulation using Olivier's script,
• Run simulation of Prather with these oxidants,
• Run simulation with Prather advective scheme (no nudging),
• See if a stable initial state for the ocean exists (with Christian and Marion).

On Thursday, 16th November

With Anne, Didier, Nicolas and Juliette.

Back to Prather's results. We are within the range of the other models in terms of lifetime. The evolution of mean annual total loss is also comparable to that of the other models, although we don't have the same spread. No significant difference with the N2O concentration à 10hPa. Shows some results on coupling with the ocean and atmosphere. A first test gives oceanic fields in good agreement with offline test.

TO DO

• Compare preindustrial runs offline and with land coupling,
• Compare difference between preindustrial runs and spinup.prather run,
• Run spinup.prather without nudging,
• Write the milestone for the end of November.

Anne has to look at the choice of advective scheme (Van Leer and/or Prather) in LMDZ and find out about interpolated oxidants.

On Thursday, 9th November

With Nicolas, Anne, Juliette and Didier.

Results of Prather's simulation. Comparison with the figures in the article (the first ten years are considered as a spin-up). Interrogation of the different models presented in the article.

TO DO

• Continue run until 2019,
• Try to make figures with a transparent background in order to better compare lifetimes and annual total loss,
• Understand the article better (how lifetimes are calculated, difference between c29 & 36 + UCI vs Oslo,..),
• Keep in mind to put INCAFLX on a svn repository (with Maureen's modif, N₂O modif and possibly do the same for NOx).

On Friday, 20th October

With Anne, Marion, Didier and Nicolas.

On the Prather's run, firsts results (3years) with N₂O atmospheric concentration at 10hPa, N₂O profiles and mean total loss. To obtain better results and because lifetime of the first year is very high, it would be preferable to start with data from the second year. For comparison, we can also run the same simulation without nudging.

TO DO

• Run a non nudged simulation,
• Play with ipslcm (same as prather, coupling with ocean...),
• Continue Prather's simulation (with Didier's files after 2014).

On Friday, 13rd October

With Nicolas, Juliette and Didier.

We found out why results were different between all our configurations. In fact, for a nudged simulation, PeriodLength? was set to 1Y but winds used were monthly. This means that we always run the simulation with the January wind (which is completely different from the August wind). By setting PeriodLength?=1M, results are better: no accumulation appears in August and the lifetime is about 164 years. It is still high but maybe changes made due to the test with Yann's files will reduce it.

We talked a little about AMA and the GA of TRACCS in January: no objection to taking part, but AMA might not be really in the scope of the work.

TO DO

• Run simulation to make comparison with Prather's article from 1985 to 2010.

On Friday, 6th October

With Didier, Juliette, Nicolas and Marion.

Simulations with Anne were carried out but I hadn't many information on them, just a plot of "vitu" vertical wind. In comparison with my runs, it seems to be different. A NMHC_AER_S configuration was set up using Yann's configuration. It doesn't seem to have this accumulation but only a few months have been computed.

TO DO

• Continue investigations on the accumulation problem,
• Run simulation in NMHC over 5 years in order to see if the accumulation appears at a given time or not.
• Also run GES over 5 years.

On Friday 22nd September

With everyone.

Juliette made a brief resume of the WP1 meeting, in which I gave a presentation about our work on N₂O. There are two milestones coming in January and May; we might run simulations to intercompare with other groups (AOD variable from 1980 to 2014, AMIP simulation in annual or monthly mean), data will be stored in ESPRI.

I recall the simulations I made during this week with some differences between each ones. Anne adds some details on how to proceed to better understand where the problem could come from and propose to have a look at source code to see which difference exist between Yann's code and mine. Didier will give me some paths towards simulations he made to see if the problem is present are not (configuration NMHC_AER_S).

TO DO

• Prepare a paper or a page on the difference and results of all those simulation,
• Send path to Anne,
• Compute burden in stratosphere only (with nco commands) + compare burden between the beginning and the end of the simulation (inter variability),
• Keep in mind Orchidee trusting.

On Wednesday 13rd September

With everyone.

Unfortunately, nudged configuration doesn't improve results; instead, it's worth because lifetime falls down to 90 years! I show some comparative plot, especially phtrate/vmrn2o/inv_o1d and temperature in order to understand the difference between the two simulations. The N₂O concentration profiles are also showed and are very different in high latitudes during the summer. After analysing the winds, there are largely different, meaning that the transport is totally different. A reason to that could be the absence of computation of ozone in the stratosphere: it seems that Zoé had a similar problem (Anne will send us documents about what she had).

TO DO

• Run simulation with ok_climoz=y in order to have computation of ozone,
• Compare simulations,
• If results are still bad, run simulation with aerosols and solar forcing.
• Compute burden in troposphere only (until 200 hPa).

From now on, meetings will occur on Fridays 2PM.

On Friday 8th September

With Didier, Juliette and Nicolas.

Simulation was done from 1990 to 2010. Lifetime is a bit high (150 years) but is decreasing to less than 140 years. Burden seems too high (1540 TgN vs 1525 TgN) while total loss are above the one of Prather (11 TgN/yr vs 13 TgN/yr). We compare the monthly N₂O loss rate with Prather's article: the amplitude is not the same and the end of year isn't equivalent also. An explanation to that high value could be the fact that the run is not nudged.

TO DO

• From 2000's, run a nudged simulation (change parameter ok_guide and add winds),
• Compare nudged and not nudged simulation (lifetime/total loss/burden/vmrn2o...),
• plot profiles of N₂O concentration,

A proposal to change the meeting time is made: possibly on Thursdays 2:30PM. Ask Anne and Marion if they have other meetings at that time.

On Thursday 31st August

With Nicolas and Didier.

Back to results obtain for year 1850. Test in 1990 doesn't work. Try to understand why there is an error in the simulation (in INCA_SRC/mksflx_p2p.f90 wrong line was put).

TO DO

• Compile and run simulation with the right line in source code,
• Compute lifetime and other variables for this run,
• Add Airmass as an output,
• Add burden and loss as outputs.

On Wednesday, 2nd August

With Juliette.

Brief remind of the last meeting. Monthly outputs are implemented, diagnostics are transformed in ferret, new lifetime from Yann oxydant is better (and decreasing from 156 years to 146years).

I have seen Patrick Brockmann to try putting diagnostics into the monitoring but unfortunately, it is not so easy.

TO DO

• Continue to implement N₂O modification into IPSLCM6.3,
• Obtain results to do the comparison with Prather article,
• Ferret transformation: pressure to altitude,
• Add outputs of burden and loss into INCA.

On Wednesday, 25th July

With Nicolas, Didier and Anne.

Mainly speak about last lifetime results and new simulations. Values are too high again.

TO DO

• Run simulation in order to compare with Prather article (1990-2005) + make comparison figures.

On Wednesday, 21st July

With Nicolas, Juliette, Anne and Didier.

Speaking of coupled model, we have to integrate DMS of Juliette in the code once N₂O is integrated and work. Try to commit orchidee modification (trusting made by Josefine) then INCA.

We had a look on vmr and o1d maps from Yann and Zoe simulations. The oxydant I use is an interpolation and we clearly see a discontinuity at a certain level. Didier says that photolyse loss occurs most in high level contrary to burden that is most at low level.

TO DO

• Run simulations with Yann oxydants and new photorate + diagnostics,
• Check if INCA modification are under GES cpp key,
• Ask Thibaut where oxydant file comes from,
• Output of invariant in monthly scale, no more daily.

On Wednesday, 12th July

With Nicolas, Juliette, Anne and Didier.

The main objective of this meeting was to understand why lifetime is so high. Didier gave me a new photorate formula and new phototable but it remains high.

Yann has some simulations with N₂O, and Zoe runs in GES configuration so I can see if lifetime/losses and burden are similar to mine.

TO DO

• Do diagnostics on Yann and Zoe simulations,
• Try to accelerate diagnostics, by using monitoring and timeseries (and transform them into ferret).
• Coupled model: test with and without ORCHIDEE modifications (interface between orch and inca).

On Wednesday, 5th July

With Nicolas, Juliette, Anne and Marion.

Mainly focus on model versions. It was set to use IPSLCM6.3 with INCA6 and ORCHIDEE3.

I always have the problem with a high lifetime (up to 150 years). We can test with new photolyse coefficient and new oxydant files.

TO DO

• Continue runs until convergence
• Time Series and packoutputs to add + use inter monitoring.

On Wednesday, 28th June

With Nicolas, Juliette, Anne and Marion.

Discussion with Anne of configurations I will use in order to transfert N₂O fluxes from pisces to inca. Two meetings was done the week before.

Preindustrial simulations show high lifetime and continuously increasing vmrn2o (no convergence). The diagnostics give a burden that is coherent with data in 39 levels but losses are too low.

TO DO

• Continue simulations with hope of convergence.
• Play with diagnostics.

On Wednesday, 21th June

With Nicolas.

HEMCO inventory: emissions are higher due to soil emissions. We can try to retrieve a percentage of agricultural emissions (40% for instance).

TO DO

• HEMCO inventory: change agricultural emissions,
• Irene: adapt code in order to read sflx files with land/ocean/misc flux,
• Run with year 1850, update orchidee + compute diagnostics with burden and loss.

On Wednesday, 14th June

With Nicolas.

TO DO

On Wednesday, 7th June

With Marion, Didier and Juliette.

TO DO

On Wednesday, 10th May

With Didier and Nicolas. Invited Thibaut and Maureen.

Discussion about simulations for ESM2025/WP1. Two different simulations: one with DMS and an other with nitrogen (NH₃ + N₂O + NO (?)). We have to know if versions of code (especially LMDZ) are similar. Thibaut is in charge of scheduling a meeting with ocean group.

We speak a little on the deliverable for WP6 (some minor correction to be made). The second GA of ESM2025 will be the next week and I have to prepare a poster: correction and finalization to do.

Box model: new results with fluxes from NMIP2: more emissions are emitted then it increases the N₂O concentration. Discussion about a pulse model in order to find the steady-state concentration. (how do they do for carbon dioxide or methane?)

TO DO

• Test the new pulse model and compare the two versions;
• Run pre industrial simulation (when Josefine files ok). First, offline with total emissions, then offline with sum of emissions (land/ocean/misc), then coupled version.
• Continue the poster.

On Wednesday, 10th May

With Didier, Juliette and Nicolas.

Discussion about deliverables of ESM2025/WP6-7. Repartition of the tasks. WP1 also needs to know which configurations they will run. It could be interesting to run simulations also with nitrogen compounds. (For example, run AMIP with transfert of NH₃ and N₂O in a NMHC configuration).

Box model: new results with fluxes from NMIP2: more emissions are emitted then it increases the N₂O concentration. Discussion about a pulse model in order to find the steady-state concentration. (how do they do for carbon dioxide or methane?)

TO DO

• Play with box model with different emissions earlier of 1850;
• Test the new pulse model and compare the two versions;
• Continue to write deliverables.

On Wednesday, 19th April

With Marion, Didier and Anne.

Code on Irene: Before the update, two versions of the code were working. Initial files and some parameters were different from one version to the other. Unfortunately, these configurations didn't work after the change of environment. Anne will see with me where this may have come from.

Marion will send a mail to organize an meeting on ocean and N₂O with Thibaut, Roland and the team.

Box model: Last results with the remarks of previous week + review of old results (Didier and Marion were not there). See attached file (pattern_boxmodel.pdf) for the different plots (with week by week updates).

TO DO

• Debug code with RedHat8.
• Run simulation for pre industrial period.

On Wednesday, 12th April

With Nicolas and Juliette. Code on Irene seems to works now. We wait for the evolution towards RedHat8 at the beginning of the next week to run a full simulation. The plan is established: run a preindustrial simulation (watch out the sst, no restart from inca), around 10 years to see whether a trends on N₂O concentration is present. Then, we can also run few years in 2000 to compare with the previous version.

Back to box model. Nicolas will send the new orchidee emissions to see the difference that can appears.

TO DO

• Test code with RedHat8.
• Continue some experiments with the box model.
• Study the convergence to a constant concentration in order to find the starting point of emissions.

On Wednesday, 5th April

With Nicolas, Juliette and Didier.

We have set a weekly meeting instead of one meeting every two weeks, as it will be easier to follow the progress of the work if someone misses a meeting.

We speak of code, because of the crashes I have on Irene when turning 79 levels. The different experiments suggested by Anne don't fix the bug. A configuration in LMDZOR works as well as a configuration from Experiments/GES_LR with only lmdz. Things are getting better but not yet solved. Nicolas asks whether I have added the orchidee patch from Maureen's code => To be checked with him.

Then, we talk about plots made thanks to the box model. I show some different shapes of emissions and the result in concentration. We see that when there is a slight slope rather than a constant emissions, the concentration is more impacted, as well as with a bump, the concentration increases much more (possibly with a small decrease afterwards).

The whole signal is important, especially the beginning of the emissions. Due to the long lifetime, the beginning of the emissions determines the evolution of the concentration (rather than the evolution at the end of the emissions). Therefor, when dummy emissions are added to the signal before 1850, the resulting concentration is affected (even if the added signal is constant).

TO DO

• Continue to investigate the bug, or try to make the code work.
• Change some experiments in the box model: plot differences in emissions and concentrations, set the baseline for emissions to 12 Tg/yr, change the white noise, play with different lifetime...

On Friday, 24th February

With Juliette, Anne, Nicolas and Didier.

Coupling with N2O between orchidee and inca is ready. Simulations are running. We need to fix initial state and some parameters. It seems that it could be better to run with 79 levels (not 39 anymore). Then, it implies to run 1 year with a rescaling from a restart of Yann and then free run.

I had modified code in order to create sflx_inputs with 3 fields for each component (land, oce and misc). I need to incorporate it in the coupled model.

Some tests were made with Maureen's code. I find the irreproducibility, I work on this problem.

TO DO

• Run coupled version with 79 levels,
• Include sflx_inputs into coupled model,
• Ask Josefine to create initial state for lmdz (limit files - 1850->1875),
• E-mail to Roland to set up a meeting all together when he will come for Jaime.

On Wednesday, 25th January

With Juliette, Anne, Nicolas, Marion and Didier.

Simulations with coupled model are good: modifications on the differents files are done. Comparison tests have been done.

Orchidee v3 has been installed and run. It exists a slight difference between Orchidee3 and Orchidee2_2 (Quest) due to new nitrogen computed fluxes.

TO DO

• Transmit nitrogen fluxes from Orchidee to Inca (using Maureen's code);
• With Didier, change names and variables in sflx_noBio: each species need to have a land, ocean and miscellaneous component (on Obelix);
• With Anne, include those changes in the INCA code to read good fields;
• Test the reproducibility of the code by doing 2*2 years or 5* 2years simulation;
• With Nicolas, create input states from Orchidee (with LMDZOR to achieve an equilibrium state): once for pre industrial period and an other for 2000's.

On Wednesday, 14th December

With Juliette, Didier and Marion.

Back to emissions reconstruction with a document of explanations, equations and figures ! It seems that:

1. the comprehension of the different methods is now better;
2. the difference between the different results is clear (or almost);
3. we have state which kind of emissions/methods we will use from 1850 until 2019.

Briefly, we talked about the black box model (Juliette wasn't here for the previous meeting). The conclusion is that either we rescale emissions in preindustrial but it will imply a too low concentration in present day or we keep these emissions but concentration are too high in preinudstrial and good for present day.

TO DO

• Finalize the document on reconstruction of emissions;
• Test a reconstruction per sector and region with an evolutive agriculture ratio (no more 0.40 for 2000 and 0.47 for 1970);
• Send a mail to Sarah and Roland in order to have explanations about the constant oceanic emissions;
• Better understand natural emissions with Nicolas.

On Wednesday, 30th November

With Nicolas, Didier and Marion.

Two main points:

1. Difference in the transfer of information into a netCDF file. Two methods are used, it is necessary to understand why they give different results. A document will be written to explain them. It is important to know where the problem comes from because we want to use good reconstruction emissions in INCA after.

2. Black box model: from emissions, use a little model in order to reconstruct concentration. This works well from 1950 to the end of the period but we get a plateau before that. This plateau is explained by the constant emissions we have for this period. By making an inverted model, it is possible to reconstruct emissions from concentration (from observations in CMIP6). It shows we need to get two or three small bumps. The absence of these emissions in ours could result in an absence of ENSO for the oceanic part or in imprecise industrial emissions for this period or a modification in land use surfaces...

TO DO

• Write a document explaining reconstruction methods and the transformation into netCDF file;
• Investigate the difference between the two methods (either by region or by region+sector), a solution could by to use a weighted factor for the sector part;
• Look on the new inventory, found by Joeran;
• Continue development on the coupled model.

On Wednesday, 16th November

With Nicolas, Juliette, Didier and Anne.

Regarding the venue of Anne, speaking about the coupled version of models. As done with Maureen, the coupling orchidee/inca is available, remains to be checked. A meeting is scheduled for the following monday between Karine and Anne to see what to do.

Invitation to participate to the "Retraite of CMC" the next two days.

Explanation and summary of the reconstruction using NMVOC as a proxy for N₂O emissions. Two methods were tested: the first one from 1970 to 2019 using the csv data, reconstruction per sector and per region; the second one from 1850 to 2019 using the proxy, reconstruction per region. It exists a slight difference in total emissions between the two.

=> Need to check where it comes from. 2 hypothesis: reconstruction from proxy uses the 2000's netCDF file as a reference and not the 1970's file (because 1970 is the year used for the reconstruction); the average of the eight factors calculated per sector doesn't correspond to the factor computed per region.

TO DO

• See the coupled configuration;
• Check where the difference between reconstructed emissions comes from;
• With Didier, black box model with land, ocean, ant emissions.

On Wednesday, 2nd November

Only with Nicolas.

Mainly focusing on N₂O reconstruction via proxy NMVOC. What I have done (using a increasing year per year ration for NMVOC and apply it on N₂O data) doesn't give good results.

New idea to use correlation between the two datasets, find the linear regression per region and use this equation to reconstruct the N₂O emissions. Looks better because the trend will be kept.

TO DO

• Continue linear regression idea (avoiding negative values in the reconstruction and discontinuities in 1970);
• Prepare slides for ESM2025 meeting (4th of November).

On Wednesday, 19th October

In person with Juliette, Didier and Nicolas.

Look on comparison of EDGAR inventory and the one of CEDS. Globally, their emissions are similar but if we look by sectors, there are different. In EDGAR, apparently, it could be better to drop fertilizers and biomass burning categories.

Use of NMVOC proxy: seems to be a good idea but have to drop some sectors already taken into account in orchidee dataset. NMVOC better than NH₃ because too related to agricultural sector, neither to CO because too related to industry, then NMVOC seem to be the less polluted by one particular sector.

TO DO

• Comparison of emissions (without soil emissions and fertilizers) and understood relationship between their subcategories;
• Have a look on geographical repartition (in order to see if patterns are similar);
• Proxy to recompute without soil emission and without domestic sector. (from 0 Tg/yr in 1750 to little in 1850);
• See with Anne to come into coupled model.

A few words on the next MERMAID meeting (December 14) with a presentation of N₂O cycle and my work (introduction with Tian graph).

On Wednesday, 5th October

In person with Didier and Nicolas.

Review of the latest results for the present day run (because Didier wasn't present at the last meeting), with:

1. Data used (BBG, NAT and ANT);
2. Results of vertical profiles and total loss for 2000 (year with an imposed rescaling at 316 ppb);
3. Results over the whole period (2000-2007) with vertical concentration, surface concentration, total loss and burden.

Show first result of countries/great regions repartition. Mainly, discussion due to a reduction in seasonality when removing soil emissions from the agricultural sector (with the help of a factor). Interrogation of which repartition country/great region to use: the EDGAR/Minx one (more precise and fine than the other two found in CEDS/Hoesly articles).

What to do before 1970? Discussion on reconstruction emissions for pre industrial period (from a proxy or with surface concentration + total loss):

• The idea of a proxy from NMVOC/CEDS could be interesting because the evolution of emissions seems to be similar between NMVOC and N₂O (by great regions and/or sectors).
• An other idea (but more atypical) could be to see the difference between burden and total loss to the atmosphere in order to determine the sources. And then, using natural and biomass burning emissions (considered as good), we can determine emissions from anthropogenic sources (globally, not sector by sector).

TO DO

• Continue to explore the CEDS inventory and the analysis by country (aggregate by great region and then use mask to reconstruct global emissions).
• Make a test: from total loss distribution and burden (of CMIP6 simulation): try to recover anthropogenic emissions.
• Use a proxy-like from NMVOC to reconstruct anthropogenic emissions from 1850 to 1970.
• Compare EDGAR emissions and CEDS emissions (global and by great regions).

The main objective of the coming weeks will be to have natural, biomass burning and anthropogenic emissions from 1850 to 1970 (in order to use them into the coupled model).

For the next ESM2025 meeting, maybe prepare some slides (on subject: Forcing data and emissions compiled and harmonized (i.e. any new forcing not available from CMIP6 - e.g. anthropogenic CH4 emissions) for use in XCA2 demonstration simulations)

On Wednesday, 21th September

In person with Juliette, Marion and Nicolas.

Presentation of recent results for the present day run, with:

1. Data used (BBG, NAT and ANT);
2. Results of vertical profiles and total loss for 2000 (year with an imposed rescaling at 316 ppb);
3. Results over the whole period (2000-2007) with vertical concentration, surface concentration, total loss and burden.

Oceanic emissions seem constant over 150 years: Marion says it is okay, because N2O concentration are not so modified with increase of temperature and so on.

Discussion about the rescaling factor: if it is too low, it means that emissions are low too; if the correction is too big, emissions are too important also.

TO DO

• Chart of CEDS without soil emissions category,
• Plot BBG emissions in same units than others graphs (meaning in Tg/yr)
• Invest if it is possible to recover input emissions from surface concentration and total loss (in order to better reconstruct emissions)

On Wednesday, 7th September

In person with Didier and Nicolas.

Now, simulation on pre industrial period is good. The pressure evolution seems to depend on temperature (sst) or anything else that does not correspond to pre industrial period (slight increasing around 2008).

Next week (or the one after), Anne will be invited to learn more about configuration/physics/radiative code and to converge step by step towards a coupled version.

TO DO

• make graph of evolution of loss (same as vmrn2o);
• run a simulation with some imposed year as 1980.
• continue to develop code for present days.
• keep in mind the work with large region (CEDS inventories, see Maureen).

On Wednesday, 24th August

In person with Didier and Nicolas. PDF support (see attached file ).

Resume of august simulations and work.

• Transformation of CEDS emissions: from 1970 to 1999, use of netCDF file 2000 with a factor per sector to create netcdf emissions. Notebooks Transform_2000nc_from1970_to1999_via_CEDScsv.ipynb and work_on_data_to_complete_sectors_with_unknows_categories.ipynb
• 2 simulations: L39.v03 as a base (pre-industrial, constrained to 285ppm), then pi_piscideeL (free, pre industrial, 40 years).

Results seem good regarding vertical pressure profile, total loss N₂O, pattern of emissions. Need to investigate more and continue.

TO DO

• Make graph of pressure for those two simulations,
• Continue piscideeL 5 years and compute mean of loss,
• Resume differences between former simulations,
• Re-create netCDF files with only manure_management (and not soil_emissions) for agricultural sector.

+ Meeting with Didier (Tuesday, 12th July)

In person with Didier.

Discussion of simulations and set-up to make.

Check N₂O loss in 1999 for v03 simulation (to be continued) and create pi_piscideeL run from it to turn 40 years (1980-2020, L means libre/free/not rescaled at 285 ppm).

Explanation of how to create ncfiles from 1970 to 1999 from 2000's geographical repartition and ratio computed with total emissions per sector (csv file). Once done, adapt ciclad's code to transform them on Incasflx 144x142 grid.

On Monday, 11th July

In person with Didier, Nicolas and Juliette.

Conversation on Workshop N₂O source in Toulouse (6-8 July) : very interesting to see different communities (observation, inversion, modellers)

First result with a decreasing total emission (from 285 ppm to 280 ppm in 20 years) : seems to be promising.

Reflection on what is needed in order to move to coupled system. Remembrance of deadlines and expectations for ESM2025 project. Land emission per year ok, Nicolas will ask to Sarah or Laurent to have same emission for ocean. After that, work with Thibault Lurton and/or Anne Cozic to adapt code.

How to reconstruct anthropologic N₂O emission between 1850 and 1970 ? Try to find some proxy (population or chemical species that exists during this period)

TO DO

• Reconstruct N₂O emission with CEDS inventory with rescaling year by year (height with total budget and spatial with year 2000)
• When all is fixed, simulation from 1970 to 2015 in order to see if the model is coherent or not (with a rescaling at ~300ppm : CMIP6).
• Find information on Japanese model (the one of Prabir Patra)
• Get more information on Pisces simulation (does it use flag or many change in order to have it? Is it possible to have emissions per year from pre-industrial to present time?)

Next

• Computation of loss rate
• Coupled model (with Anne and Thibaut)

On Wednesday, 29th June

In person with Didier and Juliette + in visio with Marion.

Explanation of the remapping problem with pisces' file. The common CDO functions doesn't work (not conservative + white columns or column with outliers).

Solution : See Germain if he has some same files (grid 294x362 and not 292x292) + in correspondence with O. Marti for a better interpolation. work in progress

TO DO

• Understand and find a solution to the remapping problem.
• Send an e-mail to Sarah Berthet, in charge of different projects on N₂O : could be seen during the N₂O workshop in Toulouse (from 6th to 8th July).
• Simulations in 39 layers with invariants outputs in two different configurations.
• On profile picture, try to invert axis + use logarithm scale.
• Check availabilities for next meetings.

+ Meeting with Didier

• See different scripts to compute the N₂O-loss and N₂O-burden.
• Show what and how to change Incasflx files in order to use a irregular grid as input (avoiding the remapping problem).

On Wednesday, 15th June

In person with Nicolas and Didier.

Waiting for the first result with the 39-layers configuration.

After the General Assembly of ESM2025, pisces emissions seems to be too high... In fact, there was a problem of units conversion (mostly because of a misunderstanding of input units (mol N rather than mol N₂O).

Nicolas asked if lightning is implemented because it seems to have an total emission between 0.5 and 2 Tg/Nyr?.

Some works will be presented during the N₂O workshop in Toulouse from July, 6th to 8th.

To DO

• Analyse the vertical profile when the 39-layers configuration of one year is done.
• + See Didier for the next steps.
• Re-compute the calculation with the good value of pisces emissions.

On Wednesday, 1st June

In person with Nicolas, Juliette, Didier. Special guest : Anne.

Special demands to Anne :

• Passage from 79 layers to 39 : is it easy and feasible ? Seems to change some configuration's files (oxidants for example).
• svn for INCA ok, if need to change orchidee's sources, ask to Nicolas (but no need for this moment).
• Make some time series for long run test (feasible with the training course),
• Add output variables that would be computing into the code (have to see Anne when it will be time).

Show results on Bouwman (with aqua and soil flux): slight difference again. Using mask 0/1 (and not fractional one) is better.

Show on-going result of CEDS inventory (from McDuffie?): problem with units, needs to calculate annual mean to correspond to article's images.

To DO

• Finish charts of CEDS inventory. Notebooks used: CEDS_charts_supplementary_data_N2O.ipynb and Charts_N2O_CEDS.ipynb.
• Schedule a meeting with Didier when 39 layers' configuration is ready.
• Make simulation with oce and soil flux for Bouwman (not with aqua !!).

On Wednesday, 18th May

In person with Juliette, Didier, Nicolas.

• Notebook Budget_inventories : total budget of N₂O emissions (in tables), differences found between outputs INCASFLX and calculations in notebook (with mask, area and change units), different charts (by latitude, by land/ocean part, with histograms...)

Using the mask with fraction:

FracMask?	- Output INCASFLX -			- Notebook computing -
(Mt/yr)	Ocean	Land	Ratio O/L	Ocean	Land	Ratio O/L
Bouwman	5.650486	11.83547	0.478	13.967	16.726	0.835
	32.3%	67.7%		45.5%	54.5%
Transcom	6.634	16.614	0.399	8.879	14.421	0.616
	28.5%	71.5%		38.1%	61.9%
Piscidee	6.432	7.107	0.905	6.579	6.989	0.941
	47.5%	52.5%		48.5%	51.5%

Using the mask with 0/1 values:

0/1 Mask	- Output INCASFLX -			- Notebook computing -
(Mt/yr)	Ocean	Land	Ratio O/L	Ocean	Land	Ratio O/L
Bouwman	5.650486	11.83547	0.478	12.797	17.899	0.714
	32.3%	67.7%		41.7%	58.3%
Transcom	6.634	16.614	0.399	8.118	15.184	0.535
	28.5%	71.5%		34.8%	65.2%
Piscidee	6.432	7.107	0.905	6.174	7.395	0.834
	47.5%	52.5%		45.5%	54.5%

Those differences can be explained by the different power that may exist between marine or continental fluxes at coasts notably. With mask, these two fluxes have the same importance so it changes total flux.

Values per latitudes:

(Tg/yr)	90-30S	30S-0	0-30N	30-90N	Total
Bouwman	5.728	7.115	9.948	8.392	31.183
Trancsom	2.714	7.432	8.762	4.857	23.765
Piscidee	1.736	4.413	4.392	3.302	13.843

Calcul file by file: From Transcom (orchidee+pisces) : (no mask + not output of sflx) -> correspond to values of table 2 thompson part 1.

	- Output INCASFLX -			- Notebook computing -
(Mt/yr)	Ocean	Land	Ratio O/L	Ocean	Land	Ratio O/L
Transcom	6.634174	16.61481	0.399293	4.23125	10.59692	0.399291

Transcom is not a pre-industrial inventory, it's actual. So, it's better to work with piscidee inventory.

To DO

• Modify Bouwman run to take only 'oce' and 'soil' variables -> relaunch notebook to compare if it's better,
• Continue modification in INCA's code.

+ Meeting with Didier (Friday, 13th May)

Informal with Didier.

Preparation to change INCA's code in Irene.

• Modify inca.card to take new NAT files and BBG from ciclad.
• Modify mksflx_p2p.F90 to put flx_n2o_ant to zero and add aflux terms.
• Modify sflx_inti.F90 to take into account BBG emissions.
• Modify set_ub_vals.F90 to rescale N₂O flux (with parallel part and constant N₂O flux at 285 ppb)

Launch different runs in order to check each step.

Next

• Change ANT in ciclad,
• Put flx_n2o_ant to 0,
• Erase the rescaling,
• Add N2O loss as an output...

On Wednesday, 4th May

In person with Juliette and Marion.

• Presentation of the Wiki page.
• Presentation of the two notebooks.
• Discussion on the different conceptions used in ocean and land communities (different grids, different units in the flux (mol/m2/s vs kg/m2/s)...).

On Analyse_3_sflx_NAT_dyn.ipynb, discussions on visualization with scale, calculation of the full emission on earth per year or per month.

To DO

• Make an analysis per month, per hemisphere, in order to get the total flux emissions for each inventory.
• Take information on where the inventories come from (especially Bouwman and Transcom).
• Draw tables which resume datas used, total emissions, models used in files...
• Compare native grid and regular grid for pisces.
• Compare values of each file used with outputs of INCASFLX.
• Compare total emissions on ocean and land with literature (ciais, ippc) and Bouwman and Trancsom inventories.

On Wednesday, 20th April

In person with Nicolas.

Discussion on:

• New page on Wiki/igcmg (this one !)
• Explanation of the created notebooks (Data_viewer and Analyse_3_sflx_NAT_dyn)
  • Data_viewer : interactive ok but add choices/texts...
  • Analyse_3_sflx_NAT_dyn : change the seasonal analysis made.
• Test with Nicolas' network to know about access to my notebooks.
• Speak about dods/orchidee to avoid downloading images on wiki.

Now, focus on N2O with long living time, but after, we can test on other compounds.

To DO

• Change colorbar with mean+/-std, add some texts on Notebooks and add units on charts
• Erase rescaling in Bouwman to have a better comparison
• Analyse the 3 datasets in a different way : global tables with sum per year, per region, per type (ocean/land) (watch out units)
• Give "procedure" and access to notebooks to all
• Compare with Auburn data (and Tian 2020)
• Look at Mendeley (or Zottero) for bibliography

Nicolas: Paths to have regional masks (from Transcom)

On Wednesday, 6th April

In Visio with Nicolas, Juliette, Didier, Marion.

Mainly, this meeting was to talk a bit more about the project and the work I(=Karine) can make before the training session (April 14th & 15th).

Discussion on:

• the first result made with INCAFLX and 3 different inventories,
• the planning we can set up (which kind of simulation...)
• the impact of N2O on chemistry (if everything is coupled),
• the certainty (or uncertainty) of inventories used (to determine an appropriate scaling),
• the work of Pierre for aquatic emissions (importance of river emissions),

Planning : Run simulation without chemistry and no rescaling, first for pre-industrial, then until now to see/understand the correction we can make on both period (a different correction for each period).

=> This have to be detailed and confirmed

Maybe run only 10 years to know where are sinks, determine some lifetime/seasonal cycle in order to calculate the scaling we may use.

Suggestions to work with the GES branch for the code (branch for coupled model with CO₂ , CH₄ and N₂O)

Meeting every 2 weeks same day, same time (ie, Wednesday at 10:00)

To DO

• Ask for writing a Wiki page on IGCMG
• Log on JupyterHub?
• Make an "experiment plan" for future simulations
• Analyse the results that I've already (seasonal analyses, comparison between inventories with Bouwman as a reference, recap origin/grid resolution...)
• Show the aquatic variable of Bouwman (and maybe some others)

Didier : emissions file for pre-industrial fires to be send to Karine