ACCESS-NRI 10 Year Roadmap - Land Modelling Input

Land Surface Working Group
,
1

Hi all,

The ACCESS-NRI Scientific Advisory Committee is developing an ACCESS Roadmap for the next 10 years. To help guide recommendations on which model development and support should be prioritised, we are starting by thinking about the most important science we want ACCESS to enable over the next decade.

As a member of the Scientific Advisory Committee I have put together a form to synthesize input from the land modelling community here: Input from Land Modelling Group for ACCESS-NRI Roadmap – Fill out form

Please provide feedback by September 7th, 2026. Input that is submitted later than this may still be considered but may not be included in the summary for a September 10 meeting in Canberra. Responses will be kept anonymous.

Please also feel free to comment on this post if you would like to start some discussion.

Best,

Alex

2

Thanks Alex for leading this important effort for the LWG. Sorry I couldn’t be there for most of discussion at the last LWG, thanks to Claire for posting the mtg notes so I could catch up.

I agree with all the points raised. I’d also like to add two additions I think are important for the next 10 yrs of ACCESS.

  1. The need for local (km-scale) seasonal and climate information, and a concern that our current generation of land models are not fit-for-purpose for this scale.
  2. A better ability to represent the impacts on climate from human interventions on the landscape (agriculture, irrigation, forestry, dams, aerosols, urban, fire management, carbon farming etc).

There is obviously a huge interest from govt, industry and public in locally relevant, nationally important weather and climate change impacts. Local interventions are something our politicians and planners have control over, very different to contributing to a global effort that can be stalled by bad actors overseas. International research groups are already undertaking global km scale simulations, and this capability for ACCESS is a stated goal of 21st Century Weather in the next 5 yrs.

These topics are also particularly relevant for land. For ocean and atmosphere modelling km-scales will resolve more processes explicitly, things should get better. But for us in land, all our previous assumptions built into our existing models (ignoring cities, lateral flow, heterogeneity, complex terrain, human interventions etc) come back to bite us, so we need to do more work.

Likewise for weather where km scales are routinely done, but in weather model deficiencies can be obscured through the good DA & AI approaches. At seasonal and climate timescales we can’t so easily hide bad physics assumptions with good data assimilation or broad empirical relations.

So I see these two points (km-scale land modelling and the ability to model the impact of human interventions on the landscape) as two critical needs requiring focus in the next 10 yrs. What worries me is the lack of investment in land model development compared with the need, here or overseas.

As I didn’t get a chance to raise these at the LWG meeting it would be great to hear what others think.

3

That’s great input Mat, thank you for sharing. I agree that both of those areas deserve more attention in a near term. The applications and relevance for the Australian community is clear for those two goals.

One question worth thinking about: In order to achieve those goals, do you think we have a modeling system that is capable of doing this – obviously with considerable effort to add processes, get better inputs/ancillaries, etc – or is there some more fundamental change needed in the model e.g. how it is configured, run and/or developed? One example could be the issue of lateral flow, which requires a rethink of how parallelisation is done.

4

I’m linking here to last year’s 12 month plan, which includes some of our highest priority items and work underway:

There is also our 2024 workshop effort to crowdsource desired developments, with a rough voting system of thumbs up on padlet: https://padlet.com/amukkola/future-cable-and-jules-model-developments-ykit7h6o3owc5a51

5

The padlet page is quite messy, but there is good community information there. That was more of a blue-sky session, so more of an appropriate snapshot for the 10-year plan that last year’s efforts (12 month plan).

To clean up I’ve asked AI to summarise these ideas and identify those with broader support. See below:

2024 Land Working Group Workshop

Well supported development ideas

Development theme Main ideas Indication of support
Improved development workflow and accessibility Faster spin-up; easier activation of new science in coupled models; shared CASA-CNP spin-up capability; integration of advances into the CABLE trunk/CABLE4. Very strong. Two items received 8 upvotes each, with additional comments. Related spin-up, integration and roadblock issues recur several times.
Human land-use and land-management impacts Crops, irrigation, fertilisation, harvesting, land-cover/use change, forest management, and urban anthropogenic heat. Strong. The broad human-modification item received 7 upvotes. Crop representation appeared again separately with 3 upvotes.
Hydrology and soil–water connectivity Lateral soil-moisture flow; river routing; water transfer between rivers and soil; groundwater links; communication between soil columns and tiles. Moderate to strong conceptual support. Lateral flow received 3 upvotes and 8 comments, the largest discussion thread. River-routing ideas recurred, although the standalone river-scheme item received 1 upvote.
Urban land-surface modelling Incorporation and completion of UCLEM in CABLE; improved urban-fraction ancillary data for JULES. Moderate. UCLEM received 3 upvotes and 1 comment; high-resolution urban data for JULES received 2 upvotes.
Dynamic vegetation structure and function Plant mortality, demography, phenology, dynamic roots, vegetation water stress, hydraulic redistribution, and Australian-specific PFTs. Moderate overall, but fragmented. Plant mortality received 3 upvotes. Phenology and Australian PFTs each generated 2 comments, despite no upvotes.
Biogeochemistry and nutrient cycling Carbon, phosphorus, soil moisture, soil microbes, and improved CASA-CNP functionality. Moderate and strategically important. The microbial biogeochemistry item received 3 upvotes; CASA-CNP spin-up received 4 upvotes and 2 comments. Phosphorus was also specifically raised in a comment.
Fire, disturbance and ecosystem mortality Fire disturbance linked to POP demography; extending BLAZE beyond CNP processes; windthrow. Moderate but uneven. Fire–demography integration received 3 upvotes, and extending BLAZE received 2. Windthrow had 1 comment but no upvotes.
Model physics and parameterisations Roughness-sublayer parameterisation; separate sunlit/shaded leaf temperatures; sensible-heat treatment; non-blackbody emissivity; latent-heat anomalies. Lower visible support. Roughness-sublayer development received 2 upvotes; most other physics suggestions received 0–1 upvotes.
Model maintenance and consolidation Remove redundant parameterisations, reconcile CABLE3/ESM1.5, and clean/test existing components before wider release. Moderate. Cleanup and reconciliation each received 2 upvotes.
Lakes and inland water bodies Represent lakes more realistically than as saturated soil. Low-to-moderate. Received 2 upvotes, with no comments.

Less commonly supported suggestions

These suggestions received little or no explicit voting support, but may still be relevant as specialist or enabling developments:

Suggestion Model Support shown
Reformulate sensible heat fluxes to account for temperature versus potential temperature CABLE 0 upvotes
Improve vegetation water-stress simulation JULES 0 upvotes
Dynamic spatial energy demand and temperature-responsive waste heat JULES 0 upvotes
Non-blackbody emissivity for soil and vegetation CABLE 0 upvotes
Evaluate or implement hydraulic redistribution CABLE 0 upvotes
Transfer dynamic phenology from CABLE-POP into ACCESS; link with JULES phenology work ACCESS–CABLE 0 upvotes, 2 comments
Windthrow or wind damage to trees CABLE 0 upvotes, 1 comment
Dynamic root systems CABLE 0 upvotes
Australian-specific plant functional types and appropriate parameters CABLE/JULES 0 upvotes, 2 comments
Two leaf temperatures for sunlit and shaded leaves CABLE 1 upvote
Investigate anomalously high latent heat around 0 W m⁻² CABLE 1 upvote
River scheme with water exchange between rivers and soil, potentially based on JULES-TRIP CABLE 1 upvote
Better lake representation CABLE 2 upvotes
Updated roughness-sublayer parameterisation CABLE 2 upvotes
Cleanup of redundant parameterisations, such as soil respiration and C4 fractions CABLE 2 upvotes

Some of the less popular ideas are nevertheless potentially important because they address model biases, physical consistency, or emerging applications. In particular, dynamic energy demand and waste heat, vegetation water stress, Australian-specific PFTs, hydraulic redistribution, and phenology integration could become more prominent if connected to specific projects or evaluation needs.

Overall interpretation

The strongest collective priorities are:

  1. Reducing barriers to model development and coupled-model integration.
  2. Representing human modification of land surfaces.
  3. Improving soil, groundwater and river connectivity.
  4. Developing more dynamic vegetation, biogeochemical and disturbance processes.

The less popular suggestions tend to be more specialised, diagnostic or process-specific. Their low vote counts should not necessarily be interpreted as low scientific value: several attracted comments, and some may function as enabling components for the more popular themes.

6

Some suggestions here (like lateral flow, IO for e.g.) might be impacted by LFRic in the online case which we don’t really have a clear picture of yet. Irrespective of LFRic particulars their are bound to be issues relating to another point raised - activating new science across platforms.

7

Thank you for that input. Mat, thanks so much for digging up those past notes and summarising the padlet info. As we all know, many discussions like these have been had before so we might as well make use of our past good work!

Jhan, yes, many science capabilities will be limited by technical capabilities, or they may even be easier to implement. The question of LFRic is a good one but an open one. The timeline could be many years before it is used with any of the ACCESS models. ESM3 will use UM13.8. So even in the medium term (~5 years) I think we should focus energy on building capability with the UM.

@Jhan can you clarify what you mean by “activating new science across platforms”?

8

I was particularly thinking about lateral flow (although it might apply elsewhere). Implementation in the offline model would not necessarily be transferable to ACCESS - even in the absence of LFRic. I appreciate that CMIP7 models will likely be relevant for ~5 years though, and perhaps we should focus our efforts here. I haven’t heard of any deviation from the 2027 timeline announced by the UKMO for the adoption of LFRic. Going back to the CAWCR days and aspirations, advancing ACCESS to operational status in NWP is not impossible. We just never had the resources. I think it would not only be a scientifically interesting/worthy pursuit but would also hopefully secure the longevity of the ACCESS model. In which case consideration of LFRic would be appropriate. Although Alex - it is highly likely that my perspective is jaded by recent events at CSIRO :slight_smile:

9

Thanks for that clarification Jhan. Yes, I can certainly why advancing ACCESS to operational status was the aspiration and worthy. I might leave it up to others with more history to chime in here and see if such an a pursuit is still a priority. I have not seen it mentioned in any of the feedback thus far but let’s use this forum as a place to discuss.

Also, I presume the adoption of LFRic here in Australia will lag the UK by quite a few years, at least for the majority of ACCESS models and applications. Although we must consider the possibility that the UM will not have ongoing support for from the UKMO.

10

If useful for us in drafting ACCESS land roadmap, I’m linking to ocean and atmosphere WG drafts:

COSIMA: Link to draft
Atmosphere: Link to draft

11

Hi all,

Thank you to everyone who has provided input thus far.

I have compiled the feedback from the recent survey, past land working group materials and other comments provided here or directly to me. With a little help from copilot I was able to synthesize these five major themes, roughly in order of priority:

Priority 1: Australian landscapes, ecosystems and climate impacts

Priority 2: Human modification and management of landscapes

Priority 3: Regional and high-resolution land modelling

Priority 4: Landscape connectivity

Priority 5: Observation-constrained Earth system modelling

In addition to this, I have compiled a number of technical capabilities that are needed to support these scientific priorities, including:

Development workflows and model accessibility

Flexible model architecture and next-generation landscape representation

Scalable computing and parallel model infrastructure

Benchmarking, Calibration and Evaluation Infrastructure

Future computing including AI and software sustainability

Finally, I have identified some cross-cutting capabilities that could and/or should be considered. THis acknowledges that many of our future priorities are not purely “land model developments”, rather they cross into other domains such as the atmosphere, ocean or cryosphere.

Community-Driven Model Development (including CMIP), Evaluation and Tuning

Model-Data Integration and Observational Constraints

High-Resolution Modelling

Cryosphere Processes

AI, Machine Learning and Hybrid Modelling

For a look at the full document including descriptions and examples of each of these topics, see this draft document:

Scientific Priorities - Land Working Group - draft v1.pdf (84.5 KB)

It is currently longer than we want so I will try to condense and put it into the format that the ACCESS-NRI SAC want.

Further comments are welcome over the next two days!

Cheers,

Alex

12

Here is the current draft of our land working group summary for first step the ACCESS 10-year roadmap development. Final comments before Thursday are welcome.

WG_priorities_2026-09-08_Land.pdf (109.8 KB)