ModelE Climate Simulations

Configurations for CMIP3 Simulations, Updates, and Issues

GISS submitted a number of different configurations to the CMIP3 model data repository at PCMDI. The configurations are as follows:

  • GISS-ER: ModelE20/Russell 4×5×L20. This uses the ModelE atmospheric code (internal version number E3), with 20 layers in the vertical, a model top at 0.1 mb and coupled to the Russell ocean model (4×5×L13). This corresponds to the rundeck E1aoM20.R in the Model E distribution.
  • GISS-EH: ModelE20/HYCOM 4×5×L20. This uses the same atmospheric model as the previous version, but coupled instead to the HYCOM ocean model (v. 0.9, 2×2×L16). This is not yet part of the public distribution.
  • GISS-AOM: (C4×3) This is a separate developmental offshoot. See here for more details.

Since the simulations were originally submitted there have been a number of updates of various fields. However, PCMDI has not placed all of these new files on their public server due to space limitations. We will endeavour to keep track of the various corrections and known problems with our simulations, as listed below, and allow access to files that are unable to be placed on the PCMDI server All such files can be found in /pub/pcmdi/. As on the PCMDI server, the directory structure is model_name/expt/run/ and all files follow the PCMDI conventions for names/structure etc. Further corrections and clarifications can be found at the PCMDI errata list.

If you require any further details on these configurations, please read the submitted papers, or contact us as described below.


Data issues:

  • All rlus/rlds files were initially incorrectly saved. Values over the ocean have however been retrieved and are available at /pub/pcmdi/modelE-R/ for all experiments.
  • rsdscs fields were calculated by sampling only clear sky boxes and cannot be compared to other models that calculated clear sky values for all boxes.
  • As described in Hansen et al. (2005; 2007a; 2007b), stratospheric ozone depletion over the period 1979 to 1997 was originally underestimated by a factor of 5/9. Simulations with the corrected ozone changes are available: 20C3M and AMIP experiments.

Code issues

  • There was a small error in the solar insolation calculation that effectively shifts the vernal equinox by a day from the intended Mar 21 hr 0. The seasonal cycle is thus off by a day but is only clearly noticeable in the insolation diagnostics (note that the models use a fixed 365 day calendar with no leap years). This affected all GISS-ER and GISS-EH/GISS-EH2 simulations.


New simulations GISS-EH2: After the submission of GISS-EH data to PCMDI, an algorithmic change was made to the ocean model which significantly improved the simulations. We have rerun all the simulations that were made with GISS-EH with the new version of the model (denoted by GISS-EH2) and we strongly encourage users to use data from GISS-EH2 instead of GISS-EH.

The specific improvement made was in the HYCOM ocean code. The density of cold sea water depends mainly on salinity and very little on temperature. Therefore diagnosing temperature from advected fields of density and salinity (the conserved variables in HYCOM) at high latitudes can lead to inaccuracies. A combination of factors related to annual mixed-layer entrainment/detrainment created biases that occasionally led to excessive ice formation in the model.

This problem was fixed when salinity was replaced by 'spiciness' during advection. For more details, see:

Unfortunately, space considerations at PCMDI has prevented the data from being available there, and so we are providing it here instead. Results from the 20C3M, PIcntrl and 1% to 2×CO2 experiments are currently available.

Other issues:

  • The corrected ozone changes during 1979 to 1997 mentioned above were used in the 20th century runs in GISS-EH2.
  • The native grid used in GISS-EH and GISS-EH2 is identical. Since the grid lines are not everywhere parallel to latitude and longitude lines as required by the IPCC data, the grid needed to be converted during post-processing. In GISS-EH2, we ended up using an output grid that is close to the native grid so that the least amount of interpolation is needed in the remapping process to preserve the information on the native grid.
  • Velocity fields in the ocean component (uo, vo) inadvertently neglected the barotropic contribution. Corrected versions are available.


No outstanding issues are known.


Please notify us about any additional issues that you come across at:

Dr. Gavin Schmidt
NASA Goddard Institute for Space Studies
2880 Broadway, New York, NY 10025 USA

+ Return to ModelE Climate Simulations Homepage