%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % README % % For Stratosphere-Mesosphere Loss Frequencies Based On % % GCClassic 14.0.0 Full-Chemistry 10-year Benchmark % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % The file is named GCC14_72LM.ch4loss.4x5.nc4 % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%% % 21 March 2023 % % % % README, *.nc4 file, and % % scripts created by % % % % Todd A. Mooring % % tmooring@fas.harvard.edu % % tmooring@alum.mit.edu % %%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 1) Middle-Atmosphere Methane Destruction % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% In GCClassic 14.1.1 and some previous versions, atmospheric methane destruction in the methane specialty simulation is supposed to occur according to three processes: 1) Reaction with OH in the troposphere 2) Reaction with Cl in the troposphere 3) Destruction at specified loss frequencies in the stratosphere and mesosphere In out-of-the-box GCClassic 14.1.1, the loss frequencies are read from a file ($ROOT/CH4/v2022-11/4x5/gmi.ch4loss.geos5_47L.4x5.nc in the out-of-the-box HEMCO_Config.rc). The loss frequencies in this file are taken from a simulation with the GMI model. In this readme we describe a new loss frequency file that is instead (largely) based on the GCClassic 14.0.0 full-chemistry 10-year benchmark simulation (and thus implicitly on that model's chemical mechanism). Inasmuch as the GMI and GCClassic 14.0.0 full-chemistry mechanisms differ, switching to this new loss frequency dataset should increase the consistency between the GEOS-Chem full-chemistry and methane specialty simulations. The new file is defined on the same spatial grid (4 x 5 deg horizontal, 72 levels) as the GCClassic 14.0.0 benchmark simulation. It was produced in three stages, described in more detail below: 1) Calculation of loss frequencies from GCClassic 14.0.0 for the bottom 59 levels (basically the troposphere and stratosphere) 2) Insertion of GMI loss frequencies in the top 13 levels (basically the mesosphere) 3) Chunking and deflation of the file %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 2) Loss Frequencies from GCClassic 14.0.0 10-year Benchmark % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% We can compute methane loss frequencies (units s-1) from four categories of output from (or input to) the GCClassic 14.0.0 10-year full-chemistry benchmark: 1) CH4 dry mixing ratio (field SpeciesConc_CH4, units mol mol-1 dry air) 2) Chemical loss of CH4 (field Loss_CH4, units molecules cm-3 s-1) 3) Dry air density (field Met_AIRDEN, units kg m-3) 4) Physical constants (molecular weight of dry air and Avogadro's number) The algorithm is obvious given the definitions of loss frequency and the above- named quantities. The GCClassic 14.0.0 simulation covers January 2010-December 2019 and the relevant outputs are available as monthly means, so all computations are applied to monthly mean data (e.g., a loss frequency is computed for January 2010 and a given grid box based on the January 2010 monthly mean mixing ratio, the January 2010 monthly mean methane loss rate, and the January 2010 monthly mean air density--all evaluated for that grid box). Although in principle this gives us interannually-varying loss frequency fields (120 values = 12 months x 10 years for each spatial location), in practice we want (and the final file of primary interest contains) a climatological seasonal cycle of loss frequencies. This is created by averaging the 10 individual-year loss frequencies available for each month and spatial location. The assignment of the year 1985 to the climatological mean file is arbitrary. The above-described computations are done using a bash script GCC14_loss_frequency_diagnosis_012023.sh primarily leveraging the netCDF Operator (NCO) toolkit. Note that this script also produces year-specific 2010-2019 loss frequency files. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 3) Use of GMI Loss Frequencies to Parameterize Mesospheric Methane Loss % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% The GCClassic 14.0.0 full-chemistry mechanism is used only in the bottom 59 levels of that model (i.e., between the surface and ~0.795 hPa). In the top 13 levels (up to the model top at 0.01 hPa--basically the mesosphere), methane destruction is instead calculated using GMI loss frequencies. Since the full- chemistry mechanism is not used in the top 13 levels, we cannot compute loss frequencies based on it at those levels. Instead, we must still resort to using GMI loss frequencies to parameterize methane loss at those levels of a methane specialty simulation. We take the relevant loss frequencies from the same GMI loss frequency field used to drive the GCClassic 14.0.0 10-year full-chemistry benchmark simulation. (Note that this is not the same file as is used to drive the out-of-the-box GCClassic 14.1.1 methane specialty simulation--instead, it has a horizontal resolution of 2 x 2.5 deg horizontal and essentially the same 72 levels as the GCClassic 14.0.0 benchmark simulation. As of 21 March 2023, this GMI file may be found on Cannon at /n/holyscratch01/external_repos/GEOS-CHEM/gcgrid/gcdata/ExtData/HEMCO/ GMI/v2015-02/gmi.clim.CH4.geos5.2x25.nc.) This GMI loss frequency information is incorporated into the top 13 levels of the *.nc4 files produced by GCC14_loss_frequency_diagnosis_012023.sh by using a MATLAB script add_GMI_loss_freqs_to_GCC14_based_files_021023.m Note that this involves horizontally averaging the GMI data from the 2 x 2.5 deg horizontal grid of gmi.clim.CH4.geos5.2x25.nc to the 4 x 5 deg horizontal grid of the GCClassic 14.0.0 benchmark output (and thus the files produced by GCC14_loss_frequency_diagnosis_012023.sh). However, no vertical interpolation or averaging is needed. gmi.clim.CH4.geos5.2x25.nc contains only a single climatological year of data, which we insert into both the climatological mean and year-specific loss frequency files from GCC14_loss_frequency_diagnosis_ 012023.sh. After the inclusion of the GMI loss frequencies for the mesosphere, the loss frequency files (both climatological mean and year-specific) contain valid values at all 72 levels. However, it is not envisioned that the loss frequencies in the troposphere will actually be used by GEOS-Chem--instead, methane loss at these levels will continue to be computed from assumed OH and Cl fields. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 4) Chunking and Deflation of Climatological Mean Loss Frequency File % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% To improve the computational performance of the model and save disk space, we have chunked and deflated the climatological mean loss frequency file using the standard nc_chunk.pl tool according to ./nc_chunk.pl GCC14_72LM.ch4loss.4x5.nc4 1 This reduced the size of GCC14_72LM.ch4loss.4x5.nc4 from 8.2M to 7.5M. The compression is lossless. The name of the final climatological mean loss frequency file we have created is GCC14_72LM.ch4loss.4x5.nc4 As of 21 March 2023, the year-specific loss frequency files have not been chunked and deflated for inclusion in a GEOS-Chem release.