NCAS Computational Modelling Services

Unified Model

Single Column Model

The Single Column Model (SCM) is a research tool which is used mainly to test/develop the UM physics code. A SCM represents a single atmospheric column at a grid-point in a General Circulation Model (GCM).

The Met Office Single Column Model is described in ​UMDP C09 on the Met Office Science Repository Service (login required).

The Met Office Single Column Model (SCM) is part of the UM code. At the Met Office it is run on local desktop PCs. This has Rose, Cylc and FCM together with GCOM and the UM shared library (shumlib). NCAS-CMS has the SCM running locally and also using the Met Office’s VM (see below).

SCM suites can be found by starting up the Rosie client running rosie go. From Rosie, search for “SCUM Template” this should provide a list of base SCM jobs. All SCM template jobs will have “SCUM Template [vn]” in the project field.

Using the Met Office’s Virtual Machine to run the SCM

The Single Column Model (SCM) can be easily run inside the Met Office’s UM Virtual Machine (VM). For full details of the VM, please see the ​Virtual Machine Guide and the associated github repository at ​https://github.com/metomi/metomi-vms

The VM should be run on a system where the user is able to install third party software. In can be run using a linux or windows host, but only the former is supported by CMS. In both cases, Virtualbox and Vagrant are required. Please see the github page for full details on these software packages. Access to MOSRS is required. The local directory where the VM was started is mounted as /vagrant inside the VM, so files cane be copied to it. It is also possible to ssh out of the VM.

Installation of the VM on Linux

  • Install virtualbox and vagrant using local package manager.

  • Clone git repo:

    git clone https://github.com/metomi/metomi-vms.git
    cd metomi-vms

  • Edit Vagrantfile.ubuntu-1804 and remove desktop from

    config.vm.provision :shell, path: "install.sh", args: "ubuntu 1804 desktop mosrs"

    so it looks like

    config.vm.provision :shell, path: "install.sh", args: "ubuntu 1804 mosrs"

    and remove the line

    v.gui = true

  • Start the VM with

    vagrant up

  • Have a coffee while it builds. When it completes, you should now be able to log onto the VM with

    vagrant ssh

    You will be asked for your MOSRS password and username (in the order).

  • While logged onto the VM, complete the installation with

    sudo apt-get install xauth

    and log out and back on again.

  • The VM is started with vagrant up, accessed by vagrant ssh, halted with vagrant halt and deleted with vagrant destroy

Installation of VM on Windows

  • Install Virtualbox and Vagrant

  • Download and install git for windows. ​https://gitforwindows.org/ Open a command window and

    git clone https://github.com/metomi/metomi-vms.git

    and cd in the metomi-vms directory

  • Build the VM with

    vagrant up

    A Virtualbox window will appear, but ignore this. Have a coffee while it builds.

  • Start the VM with

    vagrant up

    A window will appear with the VM and a terminal window open. You will be asked for your MOSRS password and username (in the order).

Preparation of the UM on the VM

Note: the following assumes that UM vn11.3 is required, if a different version is needed, replace 11.3 with that version number.

  • Install addition software packages.

    sudo install-um-extras -v 11.3

  • Install additional UM software.

    um-setup -s fcm:shumlib.x_tr@um11.3

  • Install UM data

    install-um-data

  • Install UM Metadata

    install-rose-meta

  • Install UM vn11.3

    fcm checkout fcm:um.x/trunk@vn11.3 UM11.3
    cd UM11.3
    rose stem --group=install -S CENTRAL_INSTALL=true

    This will start a rose stem suite to install various input files. The UM installation is now complete.

Running the SCM

  • Start Rosie

    rosie go &

  • Set up the MOSRS repo by clicking on Edit->Data source and selecting the Met Office’s repo.

  • Search for u-br462 and make a copy.

  • Load the job, and run the suite.