Minutes of TM5 Meeting made by F. Dentener (08.01.2003)

Place: Utrecht University

Date: 17.12.2002

Present:

Maarten Krol, Wouter Peters (IMAU)

Sander Houweling, Jos de Laat (SRON)

Frank Dentener, Peter Bergamaschi, Alexander de Meij (JRC)

Arjo Segers, Henk Eskes, Dirk Olive, Jan Fokke Meirink, Hans Cuijpers, Bram Bregman, Miranda vd Broek, Peter van Velthoven, Michiel van Weele, Ge Verver (KNMI)

Kostas Tsigaridis, Maria Kanakidou (University of Crete)

Agenda:

  1. Opening
  2. Minutes July 2002 meeting

3. Scientific presentations

  1. Standard version of TM5
  2. Model status and development
  3. Future plans and development/Action list/proposals
  1. Minutes July 2002 meeting

No comments

  1. Scientific presentations (power point presentations are available on request)

Bergamaschi: Inverse modeling results for CH4, using a preliminary version of TM5 with a zoom over Europe using Green’s function approach. Outside Europe big regions, in Europe (groups of) countries are used. 3 months of simulation

Inversion: Normalization of cost function to balance number of measurements and parameters.

Results:

-Many synoptic features are reproduced.

-Influence of different regions seen at many stations .

-Use of daily observations provide stronger constraints for the inversion than monthly values.

-Preliminary a posteriori results indicate that UNFCCC emission totals for year 2000 may be more realistic than EDGAR3.2 numbers (year 1995) for simulation period (06-08/2000).

Eskes(and Meirink, Segers, van Soest): Use of special TM3DAM version in assimilation and retrieval. One product is UV global scale retrieval ( GOME ozone columns: Profiles. Other projects: assimilation of Gome Radiances. And modelling of methane using Sciamachy methane in the shared cost action Evergreen. NO2 tropospheric retrieval. The TM3DAM special version uses a parameterised chemistry scheme. There is a paper by Eskes, will be send around. At high latitudes good predictability in extra-tropics. More problems in tropics. Re-analysis of GOME total ozone 1995-2002. Using ERA40 Brewer-Dobson is too strong, probably due to assimiliation problems.

Meirink: Sciamachy ozone columns. Compare with Gome fast delivery. Sciamachy has a bias of 20-30 DU, but this is probably to the use of old cross sections. Sciamachy problems in channels 7 and 8- dark current, dead and bad detector pixels, ice problem. At this moment retrievals are still an order of magnitude wrong. KNMI will use TMSCIA in Evergreen using 4Dvar at this moment using synthetic fields.

Van den Broek: TM5 stratospheric version, focus on HF. Paper in preparation for ACP. Different vertical definition 33 layers. Focus on horizontal resolution. Modeling of September 1999-May 2000 using OD ECMWF data and Single tracer experiments. Initisialisation is done on basis of HALOE observations. Both HF and CH4 as used as passive tracers. An additional study by van Aalst shows that the tracer fields are not sensitive to the boundary conditions and chemistry. In general different zoom resolution didn’t differ a lot, but 6-9 is too coarse too represent the vortex. TM5 performs well outside of the vortex but underestimates gradients inside. Descent rate agree quite well at > 450 K. The tracer distributions at middle and high latitudes is similar when 6x9 degrees of 2x3 degrees in the tropical region is used. A plausible reason for the discrepancies close to and inside the vortex may be the underlying meteorological wind data, and the limited vertical resolution

Olivie: Convection in ERA40 archived and off-line calculated. Archived gets higher and more intense at pressures lower than 600 hPa. Off-line downward fluxes much more different but relatively small. In tropics archived fluxes get higher, in extra-tropics rather the off-line. However regional and seasonal differences exist, so it difficult to make a general statement. Differences in upper tropospheric Rn concentrations are around 50%. Influence on NOx of different schemes is around +- 20 %. Influence on ozone in the tropics of about 20 % in tropical upper troposphere. There is a paper in preparation. It would be good to repeat the same exercise for some years in e.g. the 70s. Some studies have been made of comparing the Holtslag/Louis scheme. In general Holtslag causes less intensive mixing than Louis scheme. But to set the perspective: effects of different convection schemes are much larger.

Peters: Will work until March on the parallel version of TM5. Took over from Pourquoi. Philosophy is to distribute arrays over height and over tracers, depending on the operation. E.g.28 levels and 48 tracers.

Currently MPI is implemented, 80 % of routines rewritten. Works OK on one processor, 1 more month of work needed. Then optimizing, cosmetics, performance test. The code will be delivered to university of Crete for further optimisation on PC cluster.

De Meij: Gives overview of work using a combination of the TAPOM mesoscale model, TM5 zoom model, MISR and MODIS satellite data.Focus on Po Valey for June 2001. Prelimenary results of AOD are still a factor of 4 wrong, but this is work in progress.

  1. Standard version of TM5: implementation output and visualisation.

Krol: overview of November 2002 TM5 release. A test version with 1 pbbv exists, Radon version, two tracer verions (CO and CH4) with off-line OH. Shows IDL programs. There is an easy HIPHOP based program, which reads in meteo files, and TM5 output. Explain new data-structure.

-

  1. Model status and development:

Chemistry KPP and other issues. Sander Houweling:

Proposes to keep separate version for stratosphere and troposphere chemistry. Interface could be attained by storing fields of stratospheric run to provide boundary conditions for tropospheric version. Use concept of projects for different versions.

For tropospheric chemistry should keep different schemes which are either detailed as a quasi reference and condensed. For instance the 250-300 RACM or Kanakidou/Poisson could be a reference. However the current CBM4 should be extended with at least aceton and more biogenic VOC

Jacobs proposal of simplified heterogeneous reactions is probably reasonable.

The chemical preprocessor KPP (ROS2) is suited for development whereas EBI is still very CPU friendly and could be used for operational versions.

A central database for measurements is needed- KNMI is maintaining Brunners database (which also contains the set of ???)

NMHC emissions are at the moment lumped, where a previous TNO compilation is used to distributed NMHC over various categories. Houweling will investigate if there is a more elegant solution.

Preprocessing: Arjo Segers, Hans Cuijpers.

Special software has been developed and tested on ECMWF computer for preprocessing of ECMWF data. For more information on preprocessing see:

  • Massfluxes (pu,puv,pw); temper, humid, cloud parameters, sub, kzz, surface fields
  • Could be updated to 3 hour fields easily.
  • Cloud cover 0-1, clwc, ciwc, Cloud cover ‘overhead’ is a new field following the ECMWF definition (see manual)

Hans Cuijpers will recommend on further use of hdf, netcdf and grib data

From here on a strict naming convention for TM3-TM5 files is used following the ECMWF archiving nomenclature.

Stratospheric version: Bregman

Problems with age of air in ERA40. Problems with CFL in 60 layer version (2nd moments is too complicated). Has implemented an iteration step into z advection.

Surface fields: Dentener

About 30 surface fields are available in daily files, containing either daily average or 3 hourly varying fields. Currently its size is 20 Mb per day in compressed hdf format. Plots of surface fields are available, documentation of fields available via web page of Segers. Tests will be performed comparing ‘old’ dry deposition velocities with new ones.

TM5 documentation: Bergamaschi

Has been started, but is pending frozen version of TM5. However, well underway

(

A centralized TM5 web page is needed, which links to e.g. the meteo description of Segers, or TM5 technical details.

  1. Future plans and development/Action list/proposals
  • Action list: September 2002, Krol and Dentener made a document discussing TM5 development issues, proposed papers, possible future developments. This document which is not comprehensive and open to ‘new’ ideas is distributed separately.
  • Separate action list document added.
  • It is decided that next the full year 2001 TM5 will be processed.
  • Next meeting is scheduled Friday 6th of June.

Action list TM group, updated 17-12-02

WHO / SUBJECT / PLANNED / IN PROGRESS / FINISHED
Olivie / Model tests with vertical difusión / 28-03-02 / 12-07-02 / 17-12-02
Segers & Krol / Circulate list of preprocessing parameters / 28-03-02 / 12-07-02 / 17-12-02
Krol / TM5 standard version / 28-03-02 / 17-12-02
Bergamaschi & de Meij / TM5 documentation / 28-03-02 / 17-12-02
Cuijpers & Segers / Pre-processing ERA40 data on 2x3x60 grid / 28-03-02 / 17-12-02
Dentener & Krol & van Velthoven & Segers / Define pre-processing for cloud cover, k-diffusion, surface parameters / 12-07-02 / 17-12-02
Hans Cuijpers / Support tools
- hdf/grib/netCDF / 12-07-02 / 17-12-02
Krol / CVS for TM5? / 12-07-02 / 17-12-02
Bregman , Houweling / Parameterised stratospheric chemistry for tropospheric TM5 / 12-07-02 / 17-12-02
Houweling / Photochemical modules: what to use / 17-12-02
Van Velthoven / Cloud interpolation issues. / 17-12-02
Bregman, Meijer / Testing ERA40 winds with age of air experiments / 12-07-02 / 17-12-02