ACCESS-NRI ILAMB-workflow at NCI
What is ILAMB?
The International Land Model Benchmarking (ILAMB) benchmarking system is a python framework used to quantitatively compare a defined set of observable variables with a number of land models.
This documentation is tailored to using ILAMB on Gadi and, hence, it supplements rather than replaces the official documentation. Users are encouraged to read the ILAMB documentation and relevant tutorials.
ACCESS-NRI is releasing an NCI configuration of ILAMB under the name ILAMB-workflow.
ILAMB-workflow is the ACCESS-NRI software and data infrastructure that enables the ILAMB evaluation framework on NCI Gadi. It includes the ILAMB Python packages, a series of ILAMB outputs for ACCESS model evaluation and the ILAMB-Data collection of observational datasets. ILAMB-workflow is configured to use the existing NCI supported CMIP data collections.
Using ILAMB on Gadi
Pre-requisites
ILAMB is provided through the xp65 NCI projects on Gadi, so you need to have an NCI account and be a member of this projects to use it.
To obtain an NCI account and join NCI projects, refer to how to Set Up your NCI Account.
Depending on your needs, you may want to also join the following supported data collections:
- CMIP6:
fs38,oi10 - CMIP5:
rr3,al33
Loading the ILAMB-workflow modules
To load the the ilamb module, execute the following commands:
module use /g/data/xp65/public/modules
module load conda/access-med
For ACCESS-NRI documentation on how to run ILAMB on Gadi, visit:
To run ILAMB, you need to execute the command ilamb-run with a number of arguments/ files:
ilamb-run --config config.cfg --model_setup model_setup.txt --regions regions.txt
-
config.cfgdefines which observables and observational datasets to be compared. -
model_setup.txtdefines the paths of models that to be compared. -
regions.txtdefines the regions to be compared, e.g.,global,aust(Australia), etc.
While these files can be self-defined, ACCESS-NRI provides the necessary files and tools to set your model paths to run on Gadi. All you need to do is decide which observations and models you wish to compare.
NCI hosts replicas of the ILAMB observational data sets through the NCI project ct11 as well as a large amount of model outputs are available on Gadi, such as ACCESS model output.
For more information, refer to Observational Data Catalogue on how to find observational data on NCI and ACCESS-NRI Intake Catalog for how to find model outputs.
To learn more about how to adjust the ILAMB setup, refer to the official ILAMB documentation and relevant tutorials.
Example: CMIP6 comparisons and ACCESS ESM1.5 benchmarking
ACCESS-NRI is maintaining a collection of benchmark comparisons for the ACCESS community, such as that with Coupled Model Intercomparison Project (CMIP) data:
In the following example, the supported ACCESS Earth System Model (ESM) ACCESS-ESM1.5 is compared with two other ESM models:
- BCC ESM1 (Beijing Climate Center Earth System Model version 1)
- CanESM5 (Canadian Earth System Model version 5)
Numerous benchmark comparisons have been defined in the configuration file. The comparison of variables have been organised under different sections, such as the Hydrology Cycle.
For other variables, such as the git Gross Primary Productivity gpp, one or more datasets are available. For example, the gross primary productivity measurements of FLUXNET2015.
By clicking on a row in the table, you can expand it to see the underlying datasets used. The table's colourmap extends from best values in purple to worse data in orange.
Clicking on one of these datasets, for example CERESed4.1, will take you to an interactive and quantitative comparison page for Albedo measurements of the Clouds and the Earth’s Radiant Energy System (CERES) project:
