Skip to content

Latest commit

 

History

History
37 lines (26 loc) · 4.28 KB

README.md

File metadata and controls

37 lines (26 loc) · 4.28 KB

TRENDY v12 Experiment Configuration

This repository contains the auxiliary scripts required to run the TRENDY configuration on the RP23 NCI project on Gadi.

Usage

Copy these scripts to do the desired run directory on Gadi. Access to the HH5 project is also required to run the python landmask script. The changes needed to be made by the user are in the run_TRENDY.sh script and are:

  • Specify the location of the cable repository, by setting the cablecode variable on line 46.
  • Set the experiment to run by setting the experiment variable on line 26. The experiment names correspond to a set of internal configuration options.

The changes in this branch as compared to the trunk reflect a series of changes to the Met input routines which are currently contained in CABLE PR 290. The changes are a first pass at making the input routines generalised. The namelist options Run and MetVersion that set a series of options inside the code have been removed, and replaced with options in cru.nml which permit the same behaviour.

A the new namelist options in the cru.nml are:

  • <variable>File: CHARACTER(256), template matching the set of files for a given met variable. Includes NDep. Defaults to "None". See Met File Template for details.
  • CO2File: CHARACTER(256), location of the CO2 file. Given that the CO2 values are currently stored in a columnated text file, it does not yet require the templated format. Defaults to "None".
  • LandmaskFile: CHARACTER(256), location of the landmask file. Defaults to "None".
  • <variable>Recycle: LOGICAL, whether the given variable is recycled. Defaults to .FALSE..
  • CO2Method: CHARACTER(16), method for choosing the CO2 values. Either "Yearly" to read the current run year, or the desired year as a string. Defaults to "Yearly".
  • NDepMethod: CHARACTER(16), method for choosing the NDep values. Either "Yearly" to read the current run year, or the desired year as a string. Defaults to "Yearly".
  • LeapYears: LOGICAL, whether the Met dataset contains leap years. Defaults to .FALSE.
  • DtHrs: REAL, time step in hours. Defaults to 3.0.
  • MetRecyc: INTEGER, period for the met recycling. Defaults to 20.
  • ReadDiffFrac: LOGICAL, whether to read the diffusive fraction from file or allow CABLE to calculate it. Defaults to .TRUE..
  • RecycStartYear: INTEGER, year to start the Met recycling e.g. if MetRecycYear=1901 and MetRecyc=20, then the met forcing period will be 1901 to 1920. Defaults to 1901.
  • Directread: LOGICAL, whether to read the Met data directly from NetCDF files into vectors, or read into an array then into vectors. Defaults to .FALSE..

The run_cable.sh script has been modified to appropriately change the cru.nml namelist options for the different scenarios 0 through 3.

Met File Template

The Met files are required to be named with a certain format, that contains the start and end date of the data in the given file in YYYYMMDD format. The start and end dates are represented by <startdate> and <enddate> in the namelist option. Taking an example from the most recent TRENDY version, the file crujra.v2.4.5d.pre.1901.365d.noc.daytot.1deg.nc can be symlinked to precip-19010101-19011231.nc, and the rainFile namelist option set to precip-<startdate>-<enddate>.nc to allow it to be read by the new routines. This also handles files which span multiple years (but for now, only files which contain full years, e.g. we can't handle situations with one file containing data from 01/01/1901 to 30/06/1902 and the next from 01/07/1902 to 31/12/1903). This should allow easier substituting in and out of different Met forcing datasets.

The possible internal names for each Met variable are handled by pre-preparing a set of possible Met names stored in namelists/met_names.nml, which contains the possible NetCDF variable names to check for. To add a new entry for the possible Met names, increment the counter which sets the number of possible names, and add the new name to the array for the respective variable.