Difference between revisions of "InitMIP-Greenland"

Revision as of 08:13, 15 September 2016

Overview

Earlier large-scale Greenland ice sheet experiments e.g. those run during ice2sea and SeaRISE initiatives have shown that ice sheet initialisation can have a large effect on sea-level projections and gives rise to important uncertainties. Improving initialisation techniques is currently a field of active research, which makes it difficult to prescribe one technique as the method of choice for ISMIP6. Instead, we first propose a “Come as you are”- approach, which allows participants to contribute with their currently used model setup and initialisation technique for intercomparison (initMIP). This, we hope, allows getting modellers involved early in the ISMIP6 process and keeps the workload for participants as low as possible. Furthermore, the proposed schematic experiments may facilitate to document on-going model development. Starting early in the CMIP6 process implies relying on schematic forcing for the initiation experiments that is independent from CMIP6 AOGCM output, which will only become available later on. The initMIP-Greenland is the first in a series of ISMIP6 ice sheet model intercomparison activities and is led by Heiko Goelzer.

Goals

• Compare and evaluate the initialisation methods used in the ice sheet modelling community

• Estimate uncertainty associated with initialisation

• Get the ice sheet modelling community started with ISMIP6 activities

• Document on-going model development, as the simple experiments could be repeated with new model versions

Experimental setup

Experiments are for the large scale Greenland ice sheet and are designed to allow intercomparison between models of 1) the initial state itself and 2) the response in two schematic forward experiments:

1. init: Initialisation to present day with method of choice

2. Schematic forward experiments

2a. ctrl: Unforced control run (100 years)

2b. asmb: Prescribed schematic surface mass balance anomaly (100 years)

The two forward experiments serve to evaluate the initialisation in terms of model drift (2a, ctrl) and response to a large perturbation (2b, asmb). For 2a, the models are run forward without any anomaly forcing, such that whatever surface mass balance (SMB) was used in the initialization technique would continue unchanged. The perturbation in 2b consists of a given surface mass balance anomaly, which has to be applied relative to the initial SMB inherent to the individual initialisation technique. The SMB anomaly in 2b (the same for each model) is schematic and should not be considered as a realistic projection. The core experiment duration is set to 100 years.

Requirements for the experiments

• Participants can and are encouraged to contribute with different models and/or initialisation methods

• Models have to be able to prescribe a given SMB anomaly

• No adjustment of SMB due to geometric changes in forward experiments (i.e. no elevation – SMB feedback is allowed)

• No bedrock adjustment in forward experiment

• The choice of model input data is unconstrained to allow participants the use of their preferred model setup without modification. Modelers without preferred data set choice can have a look at the ISMIP6 wiki for possible options

• The specific year of initialization (between 1950 and 2014) is equally unconstrained to allow the use of different observational data sets that may be tied to certain time periods.

Prescribed SMB anomaly

The SMB anomaly can be obtained via the ISMIP6 ftp server (email ismip6@gmail.com to obtain the log in information), and was prepared by Heiko Goelzer. Modeling groups should use the 1km version to conservatively interpolate to their model native grid (see Appendix 1, below). Files of lower resolution (5km, 10km, and 20km) are provided for groups using the standard output grid (Bamber et al., 2001) as “native grid”. For 2b, the SMB anomaly is to be implemented as a time dependent function, that takes the form of a linear function which increases stepwise every full year (it is therefore independent of the time step in the model):

SMB(t) = SMB_initialization + SMB_anomaly * (floor (t) / 40); for 0 < t < 40 in years

SMB(t) = SMB_initialization + SMB_anomaly * 1.0; for t > 40 years

where SMB_anomaly is the anomaly provided by ISMIP6 and SMB_initialization is the SMB used for the initialization. The units of SMB_anomaly are (meter ice equivalent/year) with an assumed density of 910 kg/m^3 and 31556926 s/yr.

Specific uncertainty analysis

At a later stage and informed by the diversity and similarities of participating models, ISMIP6 will suggest further experiments to explicitly address certain aspects of uncertainty in the initialisation. It is hoped that participating groups will contribute to these additional experiments, which apply specific perturbations to the initialisations. These would take the form of repeating the experiments with systematic perturbations of the initialization choices, for example:

– Boundary conditions and other datasets

– Parameters

– Model structure

– Methods and judgments, e.g. tolerance for data mismatch or drift

- ...

References

Bamber, J. L., Layberry, R. L., and Gogineni, S.: A new ice thickness and bed data set for the Greenland ice sheet 1. Measurement, data reduction, and errors, J. Geophys. Res.-Atmos., 106, 33773–33780 (2001).

Bamber, J.L., Griggs, J.A., Hurkmans, R.T.W.L., Dowdeswell, J. A., Gogineni, S. P., Howat, I., Mouginot, J., Paden, J., Palmer, S., Rignot, E., and Steinhage, D.: A new bed elevation dataset for Greenland. Cryosphere 7, 499–510 (2013).

Appendix 1 – Output grid definition and interpolation

All 2D data is requested on a regular grid with the following description. Polar stereo-graphic projection with standard parallel at 71° N and a central meridian of 39° W (321° E) on datum WGS84. The lower left corner is at (-800000 m, -3400000 m) and the upper right at (700000 m, -600000 m). This is the same grid (Bamber et al., 2001) used to provide the SMB anomaly forcing. The output should be submitted on a resolution adapted to the resolution of the model and can be 20 km, 10 km, 5 km, 2 km or 1 km. The data will be conservatively interpolated to 1 km resolution for archiving and 5 km resolution for diagnostic processing by ISMIP6.

If interpolation is required in order to transform the SMB forcing (1km, same as Bamber et al. 2013) to your native grid, and transform your model variables to the initMIP output grid (20 km, 10 km, 5 km, 2 km, 1 km, Bamber et al. 2001), it is required that conservative interpolation is used. The motivation for using a common method for all models is to minimize model to model differences due to the choice of interpolation method. Examples of tools that have been tested by ISMIP6 members (along with tips) are available on the Regridding page.

Appendix 2 – Naming conventions, upload and model output data.

Please provide:

• one variable per file for all 2D fields

• all variables in one file for the scalar variables

• a completed readme file

A2.1 File name convention

File name convention for 2D fields:

<variable>_<IS>_<GROUP>_<MODEL>_<EXP>.nc

File name convention for scalar variables:

scalar_<IS>_<GROUP>_<MODEL>_<EXP>.nc

File name convention for readme file:

README_<IS>_<GROUP>_<MODEL>.doc

where

<variable> = netcdf variable name (e.g. lithk)

<IS> = ice sheet (AIS or GIS)

<GROUP> = group acronym (all upper case or numbers, no special characters)

<MODEL> = model acronym (all upper case or numbers, no special characters)

<EXP> = experiment name (init, ctrl or asmb)

For example, a file containing the scalar variables for the Greenland ice sheet, submitted by group “JPL” with model “ISSM” for experiment “ctrl” would be called: scalar_GIS_JPL_ISSM_ctrl.nc

If JPL repeats the experiments with a different version of the model (for example, by changing the sliding law), it could be named ISSM2, and so forth.

A2.2 Uploading your model output

Please upload your model output on the FTP server cryoftp1.gsfc.nasa.gov, and email ismip6@gmail.com for the user name and latest password. Note sftp does not work!

After log in, go to the ISMIP6/initMIP/output directory via:

ftp> cd /ISMIP6/initMIP/output

and create a directory named <GROUP> with the following sub-directory structure:

initMIP output/ <GROUP>/ <MODEL>/ init/ ctrl/ asmb/

Create additional <MODEL> directories when participating with more than one model or model version.

An example of model output files can be found in /ISMIP6/initMIP/output/ISMIP6/REF.

A2.3 Model output variables and README file

The README file is an important contribution to the initMIP submission. It may be obtained here or requested by email to ismip6-at-gmail.com

The variables requested in the table below serve to evaluate and compare the different models and initialization techniques. Some of the variables may not be applicable for your model, in which case they are to be omitted (with explanation in the README file).

We distinguish between state variables (e.g. ice thickness, temperatures and velocities) and flux variables (e.g. SMB). State variables should be given as snapshot information at the end of one year (scalars) and five year periods (2D), while flux variables are to be averaged over the respective periods.

Example model output files can be found in /ISMIP6/initMIP/output/ISMIP6/REF on the ftp server.