The initial condition file (sometimes called analysis file) is a file which must contain certain atmospheric and surface fields which the model will use to initialize the run. Therefore, the fields must cover the model grid region and the date must be the starting date of the simulation. But the resolution and grid orientation can be different than the one of the model grid.

All initial fields should be as best as possible in sync (well spun up) with the simulation.

Fields needed

Atmospheric fields

The atmospheric fields can be on model or pressure levels. Ideally, they should be on the same model levels than the model. These are the same fields that are in the pilot files for LAM simulations. Therefore, if possible, for LAM simulations they should be from the same dataset as the lateral boundary conditions data.
These are the minimum fields needed to initialize a simulation depending on the level type:

Fields on pressure levels:

• temperature (TT)
• horizontal winds (UU,VV)
• humidity (relative or specific) (HR or HU)
• geopotential (GZ)
• if available, amount of liquid and solid water in the atmosphere:
    - if P3 is used: [MPNC,MPQC,MPQR,MPNR,QTI1,QMI1,NTI1,BMI1]
    - otherwise: [QC]

Fields on model levels:

• temperature (TT)
• horizontal winds (UU,VV)
• humidity (relative or specific) (HR or HU)
• surface geopotential (GZ)
• surface pressure (P0)
• if available, amount of liquid and solid water in the atmosphere:
    - if P3 is used: [MPNC,MPQC,MPQR,MPNR,QTI1,QMI1,NTI1,BMI1]
    - otherwise: [QC]
  

Surface fields

Water/ocean:

    TM : sea surface temperature
    LG : sea ice fraction (called 'GL' in model output!)

These are the same fields as the lower boundary conditions set in 'GEM_anclima' and should therefore be from the same dataset.

Soil (land):

ISBA:

    I0 : surface and soil temperatures (2 levels)
    I1 : soil volumetric water contents (2 levels)
    I2 : soil volumetric ice contents (1 level)
    I3 : water retained on the vegetation
    I4 : water in the snow pack
    I6 : albedo of snow
    I7 : sea ice temperature
    I8 : sea ice thickness
    I9 : glaciers temperature
    SD : snow depth
    DN : snow density in kg/m3

CLASS:

same as for ISBA but:

    I0 : surface and soil temperatures (n levels)
    I1 : soil volumetric water contents (n levels)
    I2 : soil volumetric ice contents (n level)

plus the following fields:

    I5  : snow mass (SWE)
    SK : snow stored on canopy
    R2 : temperature at the base of soil water column
    Q4 : temperature of water lying on the surface
    M9 : height of water lying on the surface
    TN : ground snow temperature
    TE : canopy temperature
    GR : canopy growth factor

Lakes schemes:

FLake:

        LSF   : Lake shape factor (thermocline) (optional)
        LMLT : Lake mixed layer temperature (optional)
        LBTP : Lake bottom temperature (optional)
        LWTP : Lake water average temperature (optional)
        LIFR : Lake ice fraction
        LITH : Lake ice thickness (optional)
        LITP  : Lake ice surface temperature (optional)
        LTRN : Lake water transparency (optional)
        LMLD : Lake mixed layer thickness (optional)

Hostetler:

        LTLT : Lake top layer temperature (optional)
        LTPF : Lake water vertical temp profile (optional)
        LIFR : Lake ice fraction
        LITH : Lake ice thickness (optional)
        LITP  : Lake ice surface temperature (optional)
        LTRN : Lake water transparency (optional)
        LMLD : Lake mixed layer thickness (optional)

CSLM:

        LST : lake surface temperature
        TLAK : lake temperature profile (nl levels)
        TSDL : lake sediment temperature
        TKEL : lake mixed layer tke
        ROFL :  runoff input to lake
        HSIL : height of lake water above sill
        SNOL : snow water equivalent on lake ice
        RSNL : density of snow on lake ice
        TSNL : temp of snow on lake ice
        ASNL : alb. of snow on lake ice
        WSNL : liq.water of snow on lake ice
        DELU : current jump across lake m.l.
        LITH : total lake ice thickness
        SICE : lake snow-ice thickness

Datasets to get them from

The fields needed to initialize a simulation can get taken from different datasets. But make sure they all cover the dynamic respectively physics model grid.

For surface fields make sure they are valid for all points that contain the specific surface fraction of the model grid!!! If this is not the case these fields can get "spilled" over the invalid points.

SST & sea ice fraction

SST & sea ice fraction should come from the same dataset that will get used to drive the simulation at the lower boundary. 

Atmospheric fields

IF you are running a LAM simulation, the initial atmospheric fields should come from the same dataset that will get used to drive the simulation at the lateral boundaries. Also, both, lower and lateral BCs should come from the same dataset.

For global simulations check "Other fields" below.

Other fields

All other fields should match as good as possible the presumed state the model should have at the initial timestep.

Special care needs to get taken for fields that need a lot of time to spin up, like snow on the ground, soil temperature and moisture and lake fields.

As said above, initial conditions should be as much as possible spun up and in sync with the simulation. Fields could get taken from:

• output of a spinup simulation
• output of a similar simulation that has already run for a couple of years
• analysis data from CMC (especially for short ISBA simulations) or other institutes
• reanalysis data from ECMWF (ERA5) or other institutes