What's new in release 4.0
TM5v4 was released on October 4th, 2013. It is a frozen copy of revision #4961 (see commits 4962-4971). The following projects were released: base, chem/base, user_output, most of the levels, and budget.
New sets of data required for full chemistry model: EDGAR4.1, EDGAR 4.2, AR5, GFED3, MACC for natural emissions, HYMN/LPJG for CH4 natural emissions. These data sets replace the GEMS and GFED2 of version 3.
At interface/structure level
Note that these features do not change the results.
- PYCASSO suite of scripts to configure, compile, and run the code. Corresponding changes in rc file keys.
- MDF : universal I/O interface for netCDF and HDF.
- new faster and simpler install script
- TM6 branches for
proj/chem, and partially
user_output. Consists in a new implementation of MPI. Results are bit-identical to TM5.
- PDUMP module for timeseries output replaces old RETRO module: CF-compliant, parallel I/O.
- M7 code is now in CHEM/BASE, and not in its own sub-project (cpp /with_m7/ does the job of selecting the required source code)
Required libraries: no need for pnetcdf anymore. We rely on netcdf4 instead. To output timeseries and use restart files, netcdf-4 must be installed with parallel i/o enabled.
- bug fix that affects only zoom regions that covers the entire longitude range (pls, commit #4231)
- CFL check for zoom regions (ivdlaan, commits #4778, #4796)
- tracer slopes in convection updated consistent x,y to z (commit #4926)
- openmp bug fix (affects meteo pre-processing only, see wiki)
- option to read/write diffusion coefficient (DKG files)
- switch from full "charnock" to "Jacobson" in computation of friction velocity (mk/ajs) : abs. surface stress is now divided by air density, and we set a minum value to trap division by zero.
"Effect is only visible over sea and mixed land/sea pixels; friction velocity u* decreases with about 9% and surface roughness with about 16% . Also included a minimum value for u* to avoid division by zero in rare case that the surface stress is zero, which was observed at least once for a grid cell over Antartica (MK)." AJS Impact: "probably very small"
- Dry deposition
- updated high resistance values from 1e+5 to 1e+10
- bug fix: switch parameters from integer to real (affects chemistry only)
- CO drydep scaled from H2 drydep (Sanderson et al., 2000)
- Reduced Grid
- prevent user from using 2 contiguous 1-cell row (commits #4776, #4804)
- fix budget definition for zoom regions with resolution smaller than 1 degree (commit #4805)
- new photolysis scheme
- adds photolysis of O2 (JO2), with new O2 tracer, and new reaction rates (RCO3PO3, RCO3PO2); the scattering optical density of liquid clouds is calculated using the parameterization of Slingo (1989); the heterogeneous conversion of nitrogen pentoxide (N2O5) into nitric acid (HNO3) now uses the variable cloud droplet sizes rather than a fixed droplet size of 8μm as previously prescribed. See Williams et al. 2011 for details and impact.
- coupled to M7
- can do chemistry without convection (lightning NOX is then off)
- fixes for zoom regions (solar zenith angle, CH4 emissions nudging, latitudinal decomposition in new photolysis). See commit #4807.
- M7: can now run for more than a month without restarting
- Wet Deposition
- bug fix : use correct henry coefficients for the first time step (was random) when chemistry is on.
- bug fix : use correct henry coefficients for all time steps (was random) when chemistry is off (without_chemistry)
- scale factor relative to 100% scavenging (of HNO3) for scavenging type = 2 (ie variable factor, using henry solubility) is non-zero and computed if henry coeff > 1 (instead of 10 in base code)
- wet removal rates: in case of in-cloud scavenging, test on cloud cover has a threshold of 0.02 instead of 0.05
- fix SO2+OH : see commits #4702, 4725, 4758
- correct reading of O3du : use the exact number of levels
- do vertical regridding if needed => affect model runs with 40 and 62 levels.
- fixed : now run longer than a month have correct boundary
- Organic Nitrates bug fix
- two new chemical reactions (see commit #4376)
- xmorgntr = xmn becomes xmorgntr = xmn + 3 xmh + xmc + 3 xmo
- updated Sea Salt radii used to calculate the emitted mass. This about doubles the seas salt emissions. (Vignati et al., Atmos. Environ., 2010)
- bug fix in "calculation of look up tables rate constants/henry coefficients" : integer variables made real. Then henry(iald2) is now correct, but could have been whatever before!
- update g_n2no5 values:
before : g_n2o5_i=0.01, g_n2o5_l=0.04 now : g_n2o5_i=0.02, g_n2o5_l=0.02
- reaction_data.F90 (& chem_rates.F90)
- 3 reactions rates "for the update to no + xo2" were taken off, since they were never used. That does not change any results.
- emission_**** routines
- use combinations of EDGAR 4.1, EDGAR 4.2, AR5, GFED3, RETRO-FIRES, MACC, HYMN/LPJG inventories
- sectorial vertical distribution
- Lightning NOx: new scaling factors/in-cloud distribution (see commits #4654, #4798)
- DMS emission: replaced "nearsurface" with "surface" for vertical distribution type.
- Oxidation of SO2 and uptake of other gases in the aqueous phase (ebischeme.F90/wetS)
- CO2 now scaled to 2000 to account for annual growth of ~2ppbv/yr-1: co2=3.69e-4 instead of 3.75e-4
- new threshold to calculate Henry and dissociation equilibria, H+, sulph.: (ye(i,j,icc) > 0.02 , instead of 0.1