1 / 32

C.A.M.Brennikmeijer, S.Assonov, S.Gromov MPI for Chemistry, Mainz

Dive into the analysis of CARIBIC's flight data over the South Asian Monsoon using CO2 isotopes and Monte-Carlo simulation. Understand the complex relationships between trace gases, isotope ratios, and CO2 fluxes.

mdon
Download Presentation

C.A.M.Brennikmeijer, S.Assonov, S.Gromov MPI for Chemistry, Mainz

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. What CARIBIC sees flying overthe South Asian Monsoonthrough the prism of CO2 isotopesand Monte-Carlo Boxing C.A.M.Brennikmeijer, S.Assonov, S.Gromov MPI for Chemistry, Mainz AEA Environment Laboratories, Seibersdorf

  2. The Monte-Carlo nano-“how-to” • Largestatistics – simulate a lot! • quantitative trial and error may yield qualitative picture • MC can be used in “forward” and “reverse” modes • forward: how do results depend on parameter distributions? • reverse: which parameters lead to desired results? • MECCA has the MC simulation mode • it is “forward” only • out-of-box only probing uncertainties in reaction rates • no out-of-box facility for uncertainties in initial conditions/parameters • serial execution and output Source: Wikipedia

  3. The advanced MC engine in MESSy • This is work in progress • designed to be applicable with any sub-model • implemented for CAABA/MECCA • applied to CARIBIC observations on CO2 isotope in S. Asian Monsoon • Features • multiple tries within one run (until sufficient statistics is gathered) • generation of any number of normally/uniformly distributed parameters/boundary conditions • multiple hit targets (e.g. observed samples or experiment conditions) • multiple sets of parameters (e.g. hitting desired MRs or reactivities) • common & hit-target-wise output of desired parameters/results into netCDFs • single/parallel (using MESSy’s mo_mpi) execution, master PE and restart • highly scalable with CAABA

  4. Application: What CARIBIC sees over theSouth Asian Monsoon? • Flights 244-245, August 2008, Frankfurt ↔ Chennai, 28 samples • We focus on samples #10-19: • Beautiful latitudinal gradients in CH4, CO, O3 and δ18O(CO2), => mutual correlation! • Tight tracer-tracer relationship: CO2 and δ13C(CO2)

  5. The observations – CARIBIC isotope CO2 • Isotope CO2: • Mixing relationship in δ13C (“Keeling-plot”) • No Keeling for δ18O, but clear latitudinal gradient? • > Isotope ratios present different signals • N.B.: Opposite is seen for δ13C and δ18O in CO! CO2 uptake → CO2 respiration by vegetation ← by veg./soils (plants like (“breath” out CO2 12C over 13C) with diff. 18O/16O) NET FLUXES GROSS FLUXES

  6. What do we want to infer? • Gross regional CO2 fluxes • δ18O of local precipitation • Terrestrial 13C fractionation (BL air ↔ biosphere CO2 exchange dynamics, how plants take up CO2), seen in δ13C(CO2) and Keeling-plot • How “dirty” is the air in the Monsoon BL (e.g. O3, CO, CH4, SF6 mixing ratios) • Any better estimate on the emission ratios (e.g. w.r.t. CO) characteristic for the region/season? • How much of this air is seen by CARIBIC in the FT/UT (quantify mixing/entrainment/transport)?

  7. And how can we infer? • Using sensible assumptions on the history of air sampled

  8. And how can we infer? • Using sensible assumptions on the history of air sampled

  9. And how can we infer? • Using sensible assumptions on the history of air sampled

  10. And how can we infer? • Using sensible assumptions on the history of air sampled

  11. And how can we infer? • Sensible assumptions on the history of air sampled

  12. Difficulties • We cannot use a simpler model • It likely underparameterises the formulation (does not grasp important relationships, e.g. CO2 re-diffusion from stomata) • It never has an analytical solution • Our sample size is too small (too large?) and variable • Conditions (i.e. mixing proportions of HL, LL, BL air) change with sample • Parameter space is large • Solutions? • Use the Monte-Carlo approach • Try to “hit” every sample • Analyse in detail a selected sample with MC again

  13. The Most Sensible Model? • Compartmental kinetic exchange in CAABA/MECCA • with a very much simplified CO2 exchange with the biosphere • Hit sets: • Trace gases (TG): CO2, CO, O3, CH4, SF6 • Isotope CO2 (IC):CO2 + 13C + 18O • Known values come from CARIBIC and/or surface stations • Unknowns(*) are probed in the MC • Total parameters • 16 normally distributed • 12 uniformly distributed • Huge parameter space

  14. “Hit sets” • Used to select different successful realization scenarios

  15. Results – mixing • “Hitting” trace gases • Allows inferring air mixing proportions => Largest BL air fraction in Sample #11 (colour denotes sample no.) • Using fixed proportions • Reduces MC parameter space • Allows obtaining much moreprecise distributions (results)

  16. Results – BL trace gases • BL air composition derived from the sample #11 • Distinct MRs predicted • Allows estimating emission fluxes/ratios

  17. Results – BL isotope CO2 • BL air composition derived from sample #11 • Trace gases predict only isotopic mixing of CO2 (‘t4’ set, left) • “Hitting” isotope ratios allows to get results sensitive to biosphere-exchange processes (‘ii’ set, right)

  18. Results – BL CO2 fluxes • Relative (to advection through BL) CO2 fluxes from sample #11 • Lifetime of a CO2 molecule in uptake is 3-4x shorter than in advection • In-diffusion flux is 90-100x, retro-diffusion is 50-60x => fixation at 40x • Respiration is slow (0.2x) ?

  19. Recap • MESSy framework has an advanced MC engine • simulates chemistry in a box (or many boxes) with MECCA • applicable to any base (box) model • It allows efficiently scanning parameter space • highly scalable => huge space can be explored • multi-target => different cases in the same space • Successful application to CARIBIC observations allows to do “remote-sensing” of the Asian Monsoon BL air properties • air mixing proportions, trace gas mixing ratios • exchange with the biosphere in the BL • Application to chamber studies is exciting! • inferring reaction rates for known/proposed pathways? • quantifying any parameter (i.e. insolation, KIE, deposition)

  20. Thank you!

  21. Performance: CO2 isotopes (IC) set Try-hit statistic Sample hit “inefficiency”

  22. Performance: trace gases (TG) set Try-hit statistic Sample hit “inefficiency”

  23. What does the record of δ13C of air CO2 show? Credit: S. Assonov Mace Head, NOAA data. CO2 concentration varies due to CO2 sources/sinks. Seasonality: on local and global scale d13C co-varies with CO2. Sources/sinks have often very similar d13C signatures, e.g. photosynthesis &respiration. Difference in d13C-slopes mean variable source/sink balances. d13C observations help constraining CO2 fluxes. d13C(CO2) reflects NET CO2 fluxes.

  24. Carbon cycle: Fluxes are bidirectional Credit: S. Assonov fromICCP 2007.

  25. δ18O and CO2 exchange with leaves and soil Credit: S. Assonov Yakir et al., 2000 Bowen et al., 2003 CO2 isotopically equilibrates with water: This is exchange : d18O is not a source/sink signature; it could be NO concentration change. Different water pools (soils and leaves) have different d18O(H2O), resulting in different d18O(CO2). d18O of water in soils and leavesdepends on precipitation’ d18O, temp. & rel.humidity. Ciais et al., 1997 d18O-precipitation maps are based on extrapolation of observational data.

  26. Processes affecting the δ18O signal in CO2Credit: S. Assonov Processes & parameters forming d18O(CO2) signal : • Precipitation d18O pattern • Soil invasion • Respiration • Leaf CO2 exchange • Temperature • Evapotranspiration / air relative humidity + Transport. d18O(CO2) reflects GROSS fluxes, needs deeper understanding.

  27. What (else) can we infer? • How much of Monsoon air is seen by CARIBIC in the FT/UT (quantify mixing/entrainment/transport)?

  28. Performance: IC • The isotope CO2 (IC) probe-hit statistic

  29. Performance: TG • The trace-gas (TG) probe-hit statistic

  30. Some results, IC, sample #16

  31. Some results • Intra-sample distribution of parameters (IC)

  32. Some results • Intra-sample distribution of parameters (TG)

More Related