General information

Where can I find a general description of LS3MIP?

It is described in van den Hurk et al. (2016).

How do I subscribe to the LS3MIP mailing list?

To subscribe to the LS3MIP mailing list: Send an email to with the subject “SUBSCRIBE ls3mip ”. The firstname and lastname parameters are optional.

Which data do I have to store for LS3MIP?

A data request repository is maintained by the WGCM Infrastructure Panel (WIP), and the latest version of the Data Request can be found at The spreadsheet contains the data fields that are requested for all MIPs. The list of variables for LS3MIP can be found at ??

Time lines

When can we start the LS3MIP simulations?

The offline land surface model simulations can already be started now using the offline forcings made available (see above). The coupled model simulations generally need to duplicate a subset of the DECK or historical simulations, and it is advised to do the LS3MIP runs after completing the DECK/Historic simulations.

Data formats and portals for offline experiments

Where can I find GSWP3 forcing data for the offline experiments?

A new version of the GSWP3 forcing data set is prepared (March 2018). It now runs from 1901 to 2014. Hyungjun Kim will distribute more details via the LS3MIP? mailing list.

The main server is located in Japan.

Also, we have a mirror server at ETH, Switzerland. Please use this node from European countries.

Where can I find WFDEI forcing data for the offline experiments?

The gzipped NetCDF WATCH Forcing Data (WFD) and WFDEI files can be obtained freely from the IIASA ftp site as described on page 9 of Weedon et al. (2014) ( The WFDEI files sit within the /WFDEI directory alongside the WFD folders.

Where can I find Princeton forcing data for the offline experiments?

The PGF data are freely available at

Where can I find CRU-NCEP forcing data for the offline experiments?

CRUNCEP: It was decided to switch to CRUJRA, which is now also used in TRENDY. Data are available here.

At what resolution should I run the offline LSM?

All model groups need to generate offline model outputs as forcing to some of the LS3MIP experiments at the resolution of the native GCM grid. Moreover, some modelling groups do not have the infrastructure available to change the resolution of the offline model scheme. Therefore, the offline model data submitted to LS3MIP are expected in native model grid resolution, similar to LUMIP. The interpolation of offline forcings to the native model grid is expected to be carried out by the modelling groups.

For multi-model analyses of offline GSWP3 simulations the grid resolution of the forcing data should be followed.

Where do I store the model outputs?

There is no central repository of multi-model LSM outputs; all data are exchanged via the ESGF grid, as for all CMIP6 experiments.

Bias correction and spinup of offline land model simulations

Is there an agreed procedure to spin-up the offline land surface models?

In the LS3MIP overview paper the forcing data for 1850-1900 can be used as spin-up, and can be recycled if needed. Land use, CO2 and all other forcings should be held constant at 1850 as in the DECK 1850 coupled control simulation (piControl). However, to improve consistency with the LUMIP protocol we propose to use the TRENDY spin-up protocol, which does generate a 1850 initial condition but using atmospheric forcing data for 1901-1920. For the period 1850-1900 the same set of 1901-1920 atmospheric forcing data is cycled, but with transient (1850-1900) land use and CO2 conditions.

How are the atmospheric forcings for the simulations for future climate conditions generated?

These forcings will be derived from a selection of future climate simulations generated in the ScenarioMIP. A procedure will be followed mapping only the climate change signal from the scenario simulations onto the historical forcing.

Scenario simulations

In the original LS3MIP? description paper no decisions were taken on the choice of future scenarios for offline and coupled model simulations. A draft decision has been taken to concentrate on the scenarios SSP3-7 and SSP1-2.6 in the Tier 1 experiments, and to leave other scenarios open for Tier 2 experiments. This set of scenarios seems to be used across some other MIPs and combines the need to cover a fairly wide range while still reamining in a plausible scenario range. Comments on this decision can be communicated to the LS3MIP? coordination team. As soon as we have taken a firm decision that will be updated here.

Coordination of analyses

Which multi-model analyses will be performed with my model outputs?

The general analysis strategy is described in van den Hurk et al. (2016).

A number of analyses is already planned:

Further analyses that could/should be carried out, such as attribution of coupled model biases and model disagreement to land effects. All participants are free to design and carry out their own analyses using the outputs from LS3MIP that are made available via the ESGF.

Is there any coordination of the multi-model analyses?

LS3MIP will form an analysis coordination panel. They will frequently come together and discuss ongoing analyses, plan conference sessions or workshops, or engage the community for additional analyses. spacer