To receive email notifications when a new release is announced, please watch this topic.
ACCESS-rAM3
Release: 2.0-alpha
This an ACCESS-NRI supported release of ACCESS-rAM3, an implementation of the UK Met Office (UKMO) regional nesting suite. More information about the model is available on the ACCESS Hive Docs.
Importantly, unlike other models this is a suite that supports creating regional atmosphere/land configurations in an Australian context. The suite enables the user to choose the time and location of interest that they wish to simulate, and automatically creates the necessary fields to initialise and drive the model.
What can you expect to be able to do with this?
A user should be able to
- take this suite and run it as-is, which will run a simulation of the Lismore storm event
- modify the location of the whole simulation and the offset of the inner nest within the simulation
- modify the simulation start time
- Optionally use OSTIA sea-surface temperature and sea-ice fields for initial conditions and boundary conditions.
This is an alpha release with the intention of getting feedback from the community.
It will be fit to use and test. Known issues are listed below, but there may be some issues that have not yet been discovered and fixed. The suite may change during the alpha testing period as improvements are made. If that happens updates will made in this topic.
The documentation is incomplete at this stage. It is suitable for experienced users to run the suite, but inexperienced users may need to wait until a beta release to run the suite with confidence.
Defintions of release stages, what they mean is available on the Hive Docs:
Release stages - ACCESS-Hive Docs
For the reasons stated above this alpha release is intended for testing only and is not recommended for use in published research.
Feedback
We welcome feedback from all interested users. This is a chance to improve this proposed new release.
Please make suggestions or report problems in replies to the ACCESS-rAM3 2.0-alpha Feedback topic.
A beta release is envisaged in November. Any feedback before Monday 10th November will be considered for incorporation in the beta release.
ACCESS-NRI is interested in all feedback, positive or negative, but in particular we would like to know
- Are the instructions clear and easy to follow? Is there anything unclear, missing or incorrect?
How to use
- How to run ACCESS-rAM3
- access-ram3-config docs, which contains community contributions to ACCESS-rAM3 configuration documentation.
What’s new
Added option to use RAL3.3 (most recent version)
Updated UM atmosphere model to version 13.5
Removed anomalous rainfall spin-up in GAL9 nest
Ability to use daily-varying OSTIA (includes a new suite to generate OSTIA ancillaries)
Features
Relocatable domain (user defined)
Two-level nest, inner nest 2km resolution
Supports replace land/surface from ERA5-Land, BARRA-R2
Supports daily-varying OSTIA
Documentation (in preview mode)
Pre-compiled executable, build and deployed using ACCESS-NRI pipeline
Nesting suite
u-by395/nci_access_ram3
Default performance: 1 model day / hour (wall time), 750SU / model day
Model data output usage: 3GB / model day (default, depends on STASHPACK choices)
Driving data usage: 20GB (Automatically cleaned. Only data for most recent model run. Can be deleted after run)
Temporary work file usage: 41G (Automatically pruned. Only data for most recent model run. Can be deleted after run)
Ancillary suite (run once per nest configuration)
u-bu503/nci_access_ram3
Default performance: 70SU
Ancillary data usage: 90GB
OSTIA ancillary suite (run as needed to support nest configuration)
u-dk517
Default performance: 1 SU
Ancillary data usage: 1.2 GB
Limitations
Model calendar time is constrained by the time bounds of the forcing and boundary condition datasets.
Support is provided for 24-hour run length. Minimum run length is 24 hours.
Central location and grid box must match the grid of selected land surface initial condition data.
Known Issues
Native netCDF formatted output has been disabled
Nests must contain at least one land point with valid initial condition data. Diagnosis and work-around is covered in troubleshooting documentation.
Some combinations of start date and location can generate unstable initial conditions. There are work-arounds in the troubleshooting section of the documentation.
There are occasional intermittent unreproducible errors where RAS jobs report being stuck but have in fact failed. See diagnosis and solution in the troubleshooting section of the documentation.
Model outputs are stored in /scratch . Currently it is a user responsibility to copy this to long term storage if that is a requirement.
Large amounts of driving data are created. All but the most recent run are automatically deleted. It is a user responsibility to delete final driving data when no longer required.
The the output logs for the u-dk517 jobs are more inline with the ancillary suite. The documentation needs to be updated to reflect the newer, more reliable and faster implementation of the suite.
The GAL9 umnsaa_pb files have extra t1min_0 time profile data due to the same STASHPACK6 being used for the GAL9 and RAL3p2 nests despite different STASH time profiles being required to produce outputs. This will be looked into at a later date.
Future developments
GCM driving data is not supported in this release. This is being scoped for a possible future release.
Currently there is UM fields file output only, there is no automated post-processing to netCDF.
Credits (Latest developments only)
Based on the UKMO developed Regional Ancillary Suite (RAS) and Regional Nesting Suite (RNS)
RAS pre-filled latest possible cylc7 head-of-trunk @cbengel(Note urban ancillaries development @mlipson)
OAS suite developed by @cbengel, starting from u-cj933 suite (daily-varying ERA5 ancillaries) developed by @csu and passed on by @EmmaHoward @mlipson via the Hive. @MartinDix investigated OSTIA land/sea mask. Full suite reviewed by @mlipson.
RNS. (general) Modified head-of-trunk in preparation for migration change back-to-trunk and for making the suite useable across the whole Gadi site. Thanks to Claudio Sanchez and Anne McCabe (from the Met Office) for advice about making the suite more useable across the whole Gadi site. @cbengel migrated and tidied up u-dg768 changes to u-by395 head-of-trunk. @cbengel reviewed and updated rose/cylc vk83 independent (unless using SPACK or replace land/surface). @cbengel updated settings relating to OSTIA. @Scott reviewed suite (response still in progress).
RNS. (SPACK executable) SPACK executable was built by @paulleopardi @Aidan @harshula @TommyGatti. Softlinking to SPACK directory in u-by395 @cbengel. Changes to rose/cylc UI handling of SPACK suggested by Claudio Sanchez, modified by @cbengel with further changes suggestion by @Aidan.
RNS (GAL9 spinup issue). @coreyrobinson Raised issue GAL9 level spinup issue, @bethanwhite discussed with staff at the Met Office who described process in detail. @cbengel modified the rose/cylc suite to provide work-around.
Update to access-hive ACCESS-rAM3 content (currently PR review) @cbengel Website infrastructure @atteggiani Review of documentation @heidi
Initial content and structure of access-ram3-configs @cbengel. Website infrastructure @atteggiani
Support
Replies to this topic are disabled.
If you have specific questions about this release please reply to the ACCESS-rAM3 2.0-alpha Feedback topic.
As this is an alpha release for testing and feedback support is limited.
Resources
- ACCESS-rAM - ACCESS-Hive Docs
- How to run ACCESS-rAM3
- ACCESS-rAM3 model deployment repository
- RMED TRAC Ticket #665 to coordinate rAM3 changes
- UM support for native netCDF output [disabled in this release]