daily-spice

Enables the setup to run simulations shorter than a month. The purpose is to run very short test simulations or reanalysis-driven simulations, where the data are not yet available for the entire month. This means it is possible to set in job_settings, e.g.,
YDATE_START_ISO="1979-01-31T00:00:00Z" and
YDATE_STOP_ISO="1979-02-02T12:00:00Z"

You can install the latest version of daily-spice addone on DKRZ machine using:

# Go to SPICE folder and configure scripts:
cd SPICE_v2.3
cd configure_scripts/

# Instalation on pre-define machines:
./config.sh -s dkrz-levante -a daily-spice

After that you have to download input data:

# -- Download standard data (common for all SPICE default experiments):
cd ../data
./get_spice_rcm.sh

Finally, you need check chain folders:

cd ../chain/gcm2icon/
cd ../icon2icon

You have to be able to see new folders with the name sp001_daily-spice and sp002

Setup of sp001_daily-spice and run:

Go to the sp001_daily-spicefolder and find file job_settings:

cd ../chain/gcm2icon/sp001_daily-spice
vim job_settings

In job_settings you have to find and adapt the next parameters:

# -- Time properties --> you can play with it and use up to several hours to 30 days 
YDATE_START_ISO="1979-01-01T00:00:00Z"                # start date of simulation YYYY-MM-DDTHH:MM:SSZ
YDATE_STOP_ISO="1979-01-03T00:00:00Z"                 # end date of simulation YYYY-MM-DDTHH:MM:SSZ

# -- Resources properties: 
PROJECT_ACCOUNT=                                      # project account

Also you can check the next parameters:

# -- Path to icon folder:
ICONDIR=
# -- Notification properties:
EMAIL_ADDRESS=                                        # email address
NOTIFICATION_ADDRESS=                                 # notification email address

# -- Change only if you want to use different input data:
GCM_DATADIR=
GCM_SOILTYPE=
GCM_PREFIX=
HINCBOUND=
ICON_INPUT_OPTIONAL=
GCM_REMAP=

After that you need to save changes and run the model:

./subchain start

Setup of sp002 and run:

The second experiment, for downscaling of ICON results, can also work with daily results in this add-on. Nevertheless, the folder name remains the same. If you want to continue your experiment and use the results in your work, you will need to:

# -- Go to sp002 folder and open job_settings:
cd ../icon2icon
vim job_settings

In job_settings you have to find and adapt the next parameters:

# -- Time properties --> should be the same as in sp001
YDATE_START_ISO="1979-01-01T00:00:00Z"                # start date of simulation YYYY-MM-DDTHH:MM:SSZ
YDATE_STOP_ISO="1979-01-03T00:00:00Z"                 # end date of simulation YYYY-MM-DDTHH:MM:SSZ

EMAIL_ADDRESS=                                        # email address
NOTIFICATION_ADDRESS=                                 # notification email address
PROJECT_ACCOUNT=                                      # project account

Also you can check the next parameters:

# -- Path to icon folder:
ICONDIR=

# -- Output of sp001:
GCM_EXPID=
GCM_DATADIR=
GCM_PREFIX=
HINCBOUND=                                            # Depends on HOUT_INC parameter group 1 in sp001_daily_spice
ICON_INPUT_OPTIONAL=
GCM_REMAP=

Important notes:

There are no options to start a new run based on restart files, because we assume that you only want to use daily-spice for a short intervals and that is generally easier to recompute it.
After the first run, in case if you need to recompute icon, you can use ./subchain icon
The script arch.job.sh was simplified because this phase is not very interesting. But it still exists because it is important for sp002
There may be some problems with SAMOVAR.sh due to non-standard values. It probably makes sense to switch it off.
This add-on was tested on ERA5 data and may cause problems with ERAINT. Zo ensure it works with ERAINT or other datasets, check lines 83-157 in prep.job.sh