FILES USED BY GCM
 

INTRODUCTION
 

The COLA AGCM requires several files in order to run. The model operates in one of two modes: initial run and restart run. Each mode uses a somewhat different configuration of files while still using many files in common. Many of these files have Gaussian grid data fields which in version 2 of the AGCM are oriented as follows: the first value location is centered on the Prime (Greenwich) Meridian at the first Gaussian latitude located just north of the South Pole. The next values are located at the same Gaussian latitude and proceed eastward until the entire Gaussian latitude is encompassed. The next values are found at the next Gaussian latitude northward in the same order. This process continues to the northernmost Gaussian latitude.

Data representation varies from machine to machine. Typically IEEE standards are used, but size and byte order may vary. Most of the time big endian byte order is used because of greater portability and 4 bytes are used for integers. Real values may be represented by 4 bytes or more often 8 bytes. Output files used as input to subsequently restart the model are always written in native binary form to assure reproducibility.
 

INITIAL RUN - INPUT FILES
 

This section describes files related to INPUT used in the INITIAL RUN of GCM. The INPUT files can be divided into 4 categories: CONTROL FILES, INITIAL CONDITIONS and INITIALIZING FILES, BOUNDARY CONDITIONS FILES and INTERNAL TABLES. Whenever applicable, a subroutine name is given for referring to more details about the file in question.
 

A) CONTROL FILES
 

On some machines, the physical files are associated to the logical files in the program through script files. On these machines, the Namelist file is always read in from unit 5. The document GCM PROGRAM NAMELIST can be referred to for more details on the Namelist parameters.
 

On other machines, a filename file has been designed to be read in by optional subroutine OPNFI. This file contains the names of the physical files and some optional attributes, such as direct access I/O. This file is read in by the program, and then the listed files are FORTRAN opened to their logical counterparts, which includes the Namelist file opened from file units 15. The filename file can be edited by the user. The other files described below are included in this file.
 
 

B) INITIAL CONDITIONS AND INITIALIZING FILES
 

1) INITIAL CONDITIONS

UNFORMATTED and read in from NFIN0

This file has a complicated structure and has several records.
 

1st Record: IFDAY, TOD, IDATE, IDATEC, SIGA, SIGB

where IFDAY is INTEGER forecast day, TOD is the forecast time-of-day in seconds relative to starting time of the forecast day, IDATE is INTEGER array of dimension 4 that contains initial date (hour, month, day, year) at which initial conditions apply, IDATEC is INTEGER array of dimension 4 that contains the current date (hour, month, day, year) which will be the same as IDATE. Arrays SIGA and SIGB with 2×KMAX+1 levels each arranged top to bottom contain the hybrid vertical coordinate level information. The pure pressure based values in SIGA at the layer interfaces (odd indexes from 1 to 2×KMAX+1) and layer mid-points (even indexes from 2 to 2×KMAX) correspond to the HYAI and HYAM arrays in the AGCM. Since in this version the COLA physics cannot process hybrid coordinate levels, these are normally all set to zero except for the first top-most interface value which must be set non-zero (usually 1 Pa) when using semi-Lagrange moisture transport. The pure sigma based values in SIGB are indexed the same as SIGA and correspond to the HYBI and HYBM arrays in the AGCM. These are subsequently passed to the COLA physics.

Records 2 to JMAX+1 are individual Gaussian latitude records beginning at the southernmost latitude and progressing to the northernmost latitude with 2+4×KMAX levels or slots of data each with IMAX values indexed eastward from the Prime Meridian arranged as follows:
1st level or slot: surface pressure (Pa)
2nd level or slot: surface topography geopotential (not height) (m2/s2)
1st set of KMAX levels: (slots 3 to KMAX+2) zonal wind (m/s) indexed from the top of the atmosphere to the bottom. 
2nd set of KMAX levels: meridional wind (m/s) indexed as above
3rd set of KMAX levels: air temperature (K) indexed as above
4th set of KMAX levels: specific humidity (kg/kg) indexed as above
 

REFERENCE: Subroutines INITAL, INITCOM, HYCOEF, INIDAT and EXTIC
 

2) INITIAL SOIL TEMPERATURE AND ROUGHNESS LENGTH

UNFORMATTED and read in from NFTGZ0
 

Single read of 4 Gaussian grids: TG1, TG2, TG3, ZORL. In this model, only roughness length (ZORL) is still used.
 

REFERENCE: preparation programs intzorln.f, inttg3n.f and mktg3z.f
 

3) INITIAL SNOW MASS

UNFORMATTED and read in from NFSNW
 

Single Gaussian grid indicating initial snow mass in kg/m2.
 

NOTE:- Permanent ice points are normally set to 3000 kg/m2. Other points are set proportional to climatological surface albedo.
 

REFERENCE: preparation programs snojob.f or snowic.scr.
 

4) INITIAL SOIL WETNESS

UNFORMATTED and read in from NFSLM
 

12 months of soil moisture which are interpolated linearly to initial conditions date.
 

5) INITIAL CLOUD DISTRIBUTION (Optional File)

UNFORMATTED and read in from NFCNV0
1st Record: IFDAY,TOD,IDATE,IDATEC 

See 1st record of Initial Conditions for description.

2nd Record: CONVC,CONVT,CONVB,PRCP1,PRCP2,PRCP3,PRCPT,TOPLV,BOTLV
All Gaussian grid fields.
 

6) SiB PROGNOSTIC FIELDS (Optional File)

UNFORMATTED and read in from NFSIBI
1st Record: IFDAY,TOD,IDATE,IDATEC 

See 1st record of Initial Conditions for description.

2nd Record: 2 arrays of 2 time levels of soil temperature 

Arrays are of dimension (IMAX, 3, JMAX) where IMAX is the Gaussian longitude, 3 is the number of soil layers and JMAX is the Gaussian latitude

3rd Record: 2 Gaussian Grids of 2 time levels of ground temperature
4th Record: 2 Gaussian Grids of 2 time levels of canopy temperature
5th Record: 2 arrays of 2 time levels of soil moisture 

Arrays are of dimension (IMAX, 3, JMAX) where IMAX is the Gaussian longitude, 3 is the number of soil layers and JMAX is the Gaussian latitude

6th Record: 2 arrays of 2 time levels of liquid water storage 

Arrays are of dimension (IMAX, 2, JMAX) where IMAX is the Gaussian longitude, 2 refers to canopy and ground levels and JMAX is the Gaussian latitude

7th Record: 2 Gaussian Grids of last computed convective and large scale precipitations
8th Record: 4 Gaussian Grids:
1st Gaussian Grid - maximum mixing length
2nd Gaussian Grid - roughness
3rd & 4th Gaussian Grids - 2 time levels of sea-ice temperature
 
 

NOTE:- The data in each Gaussian latitude are left compressed to only contain land points for records 2 thru 6.
 
 

C) BOUNDARY CONDITIONS FILES
 

1) HEIGHT VARIANCE

UNFORMATTED and read in from NFVAR
 

Single Gaussian grid - units m2

Used for gravity wave drag
 

2) SEA-SURFACE TEMPERATURE

UNFORMATTED and read in from NFSST
 

Cyclical monthly SST form (using sequential data set):
 

13 Gaussian grid records - Units are KELVIN
 

1st Record: surface drag (no longer used - it is skipped)

The next 12 records refer to 12 months of height corrected SST over the ocean and a negative value over land. The values of sea-ice points, typically 270.16K, are set 1 degree below sea-ice freezing point (TSICE) . Data are treated as calendar months and can either be climatological or observed SST.
 

Example for observed SST: To allow for sufficient overlap for time interpolation an integration beginning on 1 July must have observed SST beginning in June so that records 7 to 13 are June to December of initial condition year and records 2 to 6 are January to May of the following year. If integration continues, a new data set containing April to December (records 5 to 13) of the following year and January to March (records 2 to 4) of the year after that should be used beginning 1 May of the following year. If integration still continues, data sets should be changed every ten months.
 

Continuous SST form (usually direct access weekly or monthly records) in one of two types:
 

The first type (IFSST=4) has an arbitrary number of Gaussian grid records each containing height corrected SST over the ocean and a negative value over land. As long as the time interpolated values are below TSICE (sea ice freezing), the point is treated as a sea ice point. To assure that the sea ice freezing temperature is crossed, data records with sea ice are set either to 1 degree below TSICE or as far below TSICE as the preceding or succeeding record is above TSICE.

The second type (IFSST=5) also has an arbitrary number of Gaussian grid records each containing a negative value over land, but with ordinary (non-height adjusted) SST over ocean. Here the model adjusts the SST value by height internally using the surface topography height data set (see below).

Sea ice points are determined separately by the sea ice concentration data set. This permits height adjusted values of SST to be below TSICE but still be treated as open water points. In this data set the values of SST over points with sea ice are set either to 1 degree below TSICE or weighted by the sea ice concentration so that when sea ice concentration interpolates to 0.5, the SST interpolates to TSICE.

For both types the data record time interval and starting times are indicated by namelist parameters. See IFSST, INTSST, TSICE and SSTLAG in the GCM PROGRAM NAMELIST document for usage.

3) SEA ICE CONCENTRATION

UNFORMATTED and read in from NFICE
 

This direct access data set is only read when using the second type of continuous sea surface temperature (SST) data set (IFSST=5). The interval, starting, and ending dates must match the corresponding SST data set. Points with sea ice concentration after time interpolation at or above 0.5 are assumed to be sea ice points. Units are fraction of concentration or coverage.
 

4) VEGETATION MASK

UNFORMATTED and read in from NFSIBT
 

Gaussian Grid INTEGER vegetation type over all the land points.
 

5) SOIL DATA AND MASK

UNFORMATTED and read in from NFSOIL.
 

Five Gaussian grid REAL records of soil data and soil type (5th record).
 

REFERENCE: Subroutine VEGIN
 

6) HETEROGENEOUS SPATIAL AND TEMPORAL VEGETATION VALUES

UNFORMATTED and read in from NFVEG
 

Climatological data containing 12 months of Gaussian grid data each with 3 records: leaf area index, fraction of photosynthetically active radiation (FPAR), and greenness.
 

Read in and time interpolated in subroutine GETSBC.
 

7) PLANETARY EPHEMERIDES

UNFORMATTED and read in from NFASTR
 

A complex data set containing sets of Chebyshev polynomial coefficients for each planet, the sun and the moon at 32-day intervals. See CTR#93 for further discussion. Used only when NOVAS/JPL astronomy is requested. (IAST='nova').
 

Read in subroutines INIASTRO and STATE.
 

8) ASTRONOMY DT TABLE

FORMATTED and read in from NFDELT

An optional table (TDT=.FALSE.) used only by the NOVAS/JPL astronomy (IAST='nova') to correct for time shifts caused by changes in the Earth's rotation. See CTR#93 for further discussion.
 

Read in subroutine INIASTRO.

D) INTERNAL TABLES
 

1) UNITS CONVERSION

FORMATTED - Not User Settable
 

These files are subject to expansion. The unit numbers are fixed. Six tables (files with units 34 thru 39) are used to convert model output to the units that the user wants.
 

2) AVAILABLE DIAGNOSTICS

FORMATTED - Not User Settable
 

The contents vary with model version as well as with model vertical resolution. All the information is fixed to the model version and it contains all diagnostics that the specific model version can compute. Unit number is fixed to 48. See Diagnostics System Version 2 for further discussion.
 

5) DESIRED DIAGNOSTICS

FORMATTED - User Settable
 

The contents are based on the available table used for the specific model version. Unit number is fixed to 49. See Diagnostics System Version 2 for further discussion.
 

6) VEGETATION AND SOIL PROPERTIES DATA

FORMATTED and read in from NFSIBD
 

It contains numerous parameters for each vegetation type and each soil type.
 

REFERENCE: Subroutine VEGIN
 

7) AERODYNAMIC PROPERTIES DATA

FORMATTED and read in from NFAERO
 

Contains additional parameters dependent on vegetation type.
 

REFERENCE: Subroutine VEGIN
 

INITIAL RUN - INTERMEDIATE FILE
 

This section describes the intermediate file used in the initial run of the GCM.
 

INITIAL CLOUD RADIATION STATISTICS

UNFORMATTED and written to/read in from NFCLDR
 

This file is used to save the initial cloud radiation statistics from the first radiation call for saving into the diagnostic accumulators after the accumulators have been flushed during initial start up procedures (diabatic heating, normal mode or smooth start). This permits the statistics to be retained in 2-dimensional vertical slab arrays in subsequent non-initial calls without requiring additional memory for each latitude.
 

REFERENCE: Subroutine GPHYS
 

INITIAL RUN - OUTPUT FILES
 

This section describes the output files used in the initial run of the GCM. The output files can be divided into 3 categories: FILES USED FOR SUBSEQUENT RESTART, SYNOPTIC DIAGNOSTICS, and optional SUB-SAMPLED DIAGNOSTICS.
 

A) FILES USED FOR SUBSEQUENT RESTART
 

1) ATMOSPHERIC RESTART FILE

UNFORMATTED and written out to NRG1(1,...,nnrg1). For all currently used resolutions nnrg1=1 and NRG1(1)=43.
 

A very complex data set containing both time levels of the complete atmospheric state and several intermediate product partial summations. Content and size also depend on whether running with spectral moisture transport (DOSLT=.FALSE.) or semi-Lagrange moisture transport (DOSLT=.TRUE.).

1st Record: IFDAY, TOD, IDATE, IDATEC, SIGA, SIGB, FIXMAS

The values of IFDAY, TOD, IDATE, IDATEC, SIGA, and SIGB have the same meaning as the atmospheric initial condition file (NFIN0). IFDAY, TOD, and IDATEC are set to their current values. FIXMAS is the initial dry atmospheric mass.

Most other records contain of a set of several data record groups. Each set is written out for each increasing latitude index. The set consists of: b1 derived type; b2 derived type; current time level of u, v, t and q; previous time level of u, v, t, and q. For semi-Largrange moisture transport the set also includes: lambda, phi and sigma mid-point trajectory values, and moisture and other constituent forecast values. For either moisture transport method the set ends with the two time levels of surface pressure. After all latitudes have been written, a single record set of global integral values is also written when using semi-Lagrange moisture transport. See the module PMGRID for the definitions of the b1 and b2 derived types which also depend on the moisture transport method used.

REFERENCE: Subroutines INITAL, LUNITS, RESUME, SPLITF, STEPON and WRTRS1
 

2) SiB RESTART FILE

UNFORMATTED and written out to NFSIBO
 

This file has the same structure as the INITIAL RUN INPUT FILE SiB PROGNOSTIC FIELDS (NFSIBI).

IFDAY, TOD, and IDATEC are set to their current values.

REFERENCE: Subroutine WRCOLA
 

3) CLOUD AND RADIATIVE RESTART DATA

UNFORMATTED and written out to NFCNV1
 
1st Record: IFDAY,TOD,IDATE,IDATEC as described in Initial Condition (NFCNV0) 1st record with IFDAY, TOD, and IDATEC set to their current values. 
2nd Record: cloud record (same structure of NFCNV0)
3rd Record: 9 Gaussian grid fields indicating state of the radiation and a scalar variable indicating initial total dry air mass (not currently in use)
4th Record: The upward longwave radiation accumulated at the bottom and a scalar integer with the number of time steps between longwave radiation calls. 
 

REFERENCE: Subroutine WRCOLA
 

B) SYNOPTIC DIAGNOSTICS AND SUB-SAMPLED DIAGNOSTICS

Description appears in Diagnostics System Version 2 document.
 

RESTART RUN - INPUT FILES

This section describes files related to INPUT used in the RESTART RUN of GCM. The INPUT files can be divided into 4 categories: CONTROL FILES, BOUNDARY CONDITIONS FILES and INTERNAL TABLES, INITIAL FILES. Whenever applicable, a subroutine name is given for referring to more details about the file in question.
 
 

A) CONTROL FILES

These files are the same described in the INITIAL RUN.
 
 

B) BOUNDARY CONDITIONS FILES

These files are the same described in the INITIAL RUN.
 
 

C) INTERNAL TABLES

These tables are the same described in the INITIAL RUN.
 
 

D) INITIAL FILES
 

1) ATMOSPHERIC RESTART FILE

UNFORMATTED and read in from NRG1(1,...,nnrg1). For currently used resolutions, nnrg1=1 and NRG1(1)=41. This is the same file described in INITIAL RUN - OUTPUT FILES: FILES FOR SUBSEQUENT RESTART.
 

2) SiB RESTART FILE

UNFORMATTED and read in from unit NFSIBI. This has the same file described previously in the INITIAL RUN INPUT FILES SiB PROGNOSTIC FIELDS.
 

3) CLOUD AND RADIATION RESTART FILE

UNFORMATTED and read in from unit NFCNV0. It has the same structure as CLOUD AND RADIATION RESTART FILE (unit NFCNV1) described in the INITIAL RUN - OUTPUT FILES: FILES FOR SUBSEQUENT RESTART.
 
 

RESTART RUN - INTERMEDIATE FILE

The intermediate file is not used for RESTART RUN..
 
 

RESTART RUN - OUTPUT FILES

All these files are the same described in the INITIAL RUN OUTPUT FILES.