Climate Science Initiative – Frequently Asked Questions

Answers to frequently asked questions about the Climate Science Initiative (CSI).
Last updated:

What is the Climate Science Initiative and why is it important?

Climate change poses significant challenges for communities in Western Australia (WA) who are experiencing more extreme weather events such as tropical storms, floods and bushfires. The State Government is supporting businesses, communities and local governments to understand the future climate and adapt to the impacts of climate change.

The CSI is developing climate projections for WA at 20-km and 4-km resolution in partnership with the NSW Government’s Department of Climate Change, Energy, Environment and Water, Murdoch University and WA’s Pawsey Supercomputing Research Centre.

The climate projections will represent our best understanding of WA’s future climate to 2100. The projections will support government agencies, businesses and individuals to prepare for events such as storms, floods, bushfires, and heatwaves, and protect WA’s unique biodiversity.

How are climate projections developed and why do we need them?

A global climate model (GCM) is a computer simulation of the Earth’s climate system based on real data. GCMs represent processes and interactions that drive the Earth’s climate and help us to understand changes resulting from activities such as burning of fossil fuels. The coarse resolution of GCMs (from 100 to 250 km) limits their ability to examine the impacts of climate change on local regions or towns.

The CSI will give us a better understanding of climate change at a local level by producing climate projections at smaller scales (20 km and 4 km resolution). At these scales, regional influences like topography, land use and coastlines can be better understood, as well as extreme events like storms. The figures below illustrate the improved resolution with a 4 km grid cell size.

A map of a region Description automatically generated

Figure 1 Example of land use in a global climate model demonstrating the resolution difference with grid cell size

Figure 2 Example of topography in a global climate model, which also shows the need for downscaled regional models

Modelling at this scale has never been carried out for WA. It will represent our best understanding of how the climate will change over the next 75 years, using a range of important climate variables, including rainfall, temperature, humidity, and air/wind speed.

What will climate projections tell us about the future climate?

Climate projections will help government and industry manage the risks associated with climate change by providing robust future climate scenarios.

The projections will provide an overview of how conditions will change on average in a region over the coming decades (for example, the likely change in daily maximum summer temperatures in Joondalup by 2050). The projections will also provide vital information on climate extremes including the number of very hot days, severe fire danger days, drought duration and intensity of rainfall events.

The CSI will help answer questions like:

  • What risks does climate change pose to specific regions?
  • What impact can we expect on rainfall and regional water supplies?
  • How will cyclones and tropical storm patterns change?
  • What will be the change in fire risk in the southwest?

It is important to note that climate projections involve uncertainties due to the complexity of the Earth’s climate system and future greenhouse gas levels. CSI will synthesise and present the best available scientific information to support decision making.

Where can I find credible information about CMIP 6 and climate modelling?

CMIP6 models represent a new generation of climate models. For more information see our explainer on the latest in climate modelling: Everything you need to know about the latest in climate modelling factsheet. 

For further information about CMIP6 and climate modelling in general: 

Who should use the Climate Science Initiative projections?

Climate projections are used by a wide range of users, including scientists, policymakers, government agencies, planners, engineers, businesses, non-profit organisations, and the public.
 
These users and their information needs are summarised in CSI’s infographic Who are the users of CSI data? | Western Australian Government.

When will Western Australian climate projections become available and what is coming next from the CSI?

Statewide temperature and rainfall climate projections for the whole of WA at a grid resolution of 20 km are available for technical users from the Shared Location Information Platform (SLIP) | Data WA.  

The projection simulations for the North West, covering the Pilbara and Kimberley regions, are anticipated to be available in late 2026. An online climate viewer map is in development and will visualise the projections for non-technical users. It is expected to be available by the end of 2026. 

Regional summaries are also under development for South West WA using the new 4 km projections. These will provide simple, accessible overviews of projected temperature and rainfall changes in various regions and will make climate projections more accessible for communities, businesses and other stakeholders.   

Producing local climate projections is complex, highly technical work and is being undertaken for the first time in WA. The cutting-edge nature of this work involves solving tens of thousands of equations at every single grid point and timestep which represents WA’s regional climate.

The department will provide regular updates on delivery timeframes for climate projections as the project progresses.

What global climate models are being used by the Climate Science Initiative and how were they chosen?

Many Global Climate Models (GCMs) exist, each with strengths and weaknesses in projecting various weather parameters such as temperature and rainfall, for different areas of the globe. They are developed by teams around the world, particularly in the Northern Hemisphere. This means it is important to carefully evaluate and select models that are appropriate to use in WA. The GCMs used by the CSI have been carefully chosen to ensure they best represent the climate of WA, are statistically independent from one another and consider the climate diversity of Australia. Models chosen described are in Figure 4 below.
 
Five global climate models used in the Climate Science Initiative
 Figure 3 Five global climate models used in the Climate Science Initiative infographic
 
To provide locally relevant information, the Climate Science Initiative applies a regional climate model called Weather Research & Forecasting Model (WRF) system to dynamically downscale GCMs. The WRF modelling system is an established and regularly updated regional climate modelling system that simulates the climate at finer scales than GCMs. CSI uses the current version WRF4.1.2.
 
Rigorous and peer-reviewed analysis has been undertaken to identify the best-performing models. The GCMs selection methodology is consistent with the NSW Government Department of Planning and Environment NARCliM project.
 
Further details are provided in the following reports produced by NSW DCCEEW:

Will the Climate Science Initiative model sea level rise and coastal erosion risk?

The Climate Science Initiative is not modelling sea level rise. The CSI climate projections are most useful for terrestrial environments and land-based applications rather than marine or coastal planning.
 
However, model outputs from the CSI can be used by expert users in nearshore wave models to understand coastal inundation and erosion. Resources to assist with sea level rise and marine heatwaves are available here:’ A guide to climate science resources for Western Australia | Western Australian Government (www.wa.gov.au).
 
For additional advice, see the updated Climate Change Considerations chapter of the Australian Rainfall and Runoff: A Guide to Flood Estimation (ARR) released by the Department of Climate Change, Energy, the Environment and Water in partnership with Engineers Australia.

What climate projections should I use while the CSI projections are being developed?

There is a wide variety of climate projection datasets currently available, and the right choice depends on the user's needs. Visit our guide to climate science resources for Western Australia for a list of resources currently available, including consideration of other sources of climate projection data.

Which emissions scenario should I consider?

For policy and climate risk assessments, it’s useful to consider at least two emissions scenarios to provide a picture of a range of possible future climates. A climate risk assessment methodology can guide the selection of the most appropriate scenarios to use for a particular purpose, or an organisation may have a policy position that specifies which scenarios to use. A common practice is to consider at least two alternative climate futures.
 
For most purposes, this would include one scenario representing the lower range of future emissions (e.g. SSP1-2.6 or SSP2-4.5) and one representing a high-range emissions trajectory (e.g. SSP3-7.0). These scenarios can be used to comply with sustainability standards set by the Australian Accounting Standards Board under the Corporations Act 2001. The Act requires certain entities to disclose information about its climate resilience, as assessed under at least two scenarios. The mandated scenarios are set out in the Climate Change Act 2022 as follows:
 
  • Increase in global average temperature of 1.5°C above pre-industrial levels; and
  • Increase in global average temperature well exceeding 2°C above pre-industrial levels (meaning an increase of 2.5°C or higher).
     
While the use of specific SSPs is not mandated, the scenarios above broadly align with projections in scope for the Climate Science Initiative.
 
It may be appropriate in some projects, especially those investigating climate extremes or asset-level risk assessments, to explore a high-emissions or ‘worst case’ scenario available. Analysis of a high emissions scenario enables consideration of severe future climate impacts, enabling stress testing of the sector to inform adaptation action development.
 
The developers of the Shared Socioeconomic Pathways do not estimate the relative likelihood of any scenario. Since any future scenario is plausible, it is best to consider multiple scenarios where possible. Currently, SSP2-4.5 is the closest reflection of present-day climate mitigation and adaptation actions (UNEP, 2022).
 
The WA Government Climate Risk Management Guide (interim) recommends state agencies use two climate scenarios for first-pass physical climate risk assessments, RCP 4.5 (intermediate emissions) and RCP 8.5 (high emissions). Previously within the climate science community, RCP8.5 was considered to be a good representation of a high range emissions scenario for the previous CMIP5 suite of models that could be used to inform the upper bounds of possible climate change for policy development and risk assessments. However, changes in baselines, the passing of time, more recent data and an improved understanding of how the main drivers of future emissions are likely to change over time mean that SSP3-7.0 is now considered to be good representation of a high-end emissions scenario within the CMIP6 suite of models (i.e. because emissions higher than described in SSP3-7.0 are now considered unlikely). For more information on RCPs and SSPs please see the CSI’s infographic on this transition or our climate modelling factsheet.
 
Each new generation of climate models represents an increased understanding of climate processes. In general, no source of information is right or wrong, and new generations of modelling do not mean findings from earlier generation modelling are not useful or plausible anymore.

What are the differences between the newer CMIP6 models compared to the previous CMIP5 models?

The Coupled Model Intercomparison Project – delivered in phases (e.g., CMIP5 and CMIP6), is coordinated by the World Climate Research Programme. CMIP5 models informed the IPCC’s Fifth Assessment Report (AR5). CMIP5 models considered several different trajectories for future changes in the concentration of atmospheric GHGs described using Representative Concentration Pathways (RCPs).
 
Coupled Model Intercomparison Project Phase 6 (CMIP6) models are the latest generation of global climate models used for climate research and projections. The results from CMIP6 models feature in the latest Sixth Assessment Report (AR6) from the IPCC. CMIP6 utilized the new set of emissions scenarios called Shared Socioeconomic Pathways (SSPs). Strengths of CMIP6 models include: 
 
  • improved representation of key processes
  • higher resolution, and
  • the inclusion of additional Earth system components.
     
Until the full suite of CMIP6 projections are available, users can still use CMIP5 projections where required, and prepare to transition to, or supplement with, CMIP6 data as it becomes available. The Guide to future climate projections for water management in Western Australia proposes an approach for how to determine whether a previous assessment using earlier projections is still fit for purpose. 
 
It’s important to note that there are only incremental differences between these generations, and broad trends remain largely the same. For first pass climate risk assessments and adaptation planning, CMIP6 modelled results are unlikely to offer significant differences to CMIP5 results and impact previous risk classifications or adaptation decisions. This is due to the large consistency between CMIP5 and CMIP6 modelled results, because climate change impacts are usually evaluated at a broad scale where incremental improvements in model resolution and physics (present in CMIP6) do not substantially change the projections and associated risk classifications already assessed for CMIP5. The general magnitude of projected changes remains sufficiently similar across CMIP5 and CMIP6, and as such previous or current analysis based on CMIP5 models remains robust and can confidently be used in climate risk assessments.

Why is the CSI modelling different scenarios? What emissions scenarios have been used to produce the projections and why?

Global temperatures are currently predicted to warm by between 1.8°C by 2100 for a low greenhouse gas emissions scenario, and 3.6°C by 2100 for a high greenhouse gas emissions scenario. The CSI will help us to understand what these global trends will mean for WA.
 
Emissions scenarios used in the CSI are based on the Shared Socioeconomic Pathways (SSPs) described in the Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Report (AR6).
 
The CSI is modelling three future emissions scenarios to understand what may happen in the future (Table 1):
  • a low emissions scenario where temperatures increase by about 1.8°C by 2100 (SSP1-2.6). This scenario envisions a low-emissions future with a global transition towards sustainability, including stringent environmental policies and sustainable practices, leading to lower global greenhouse gas emissions.
  • a medium emissions scenario where temperatures increase by about 2.7°C by 2100 (SSP2-4.5). This scenario is an intermediate greenhouse gas emissions scenario where global efforts to reduce emissions are moderately successful. Some regions may adopt sustainable practices and technologies, while others continue with higher emissions, leading to moderate climate changes.
  • a high emissions scenario where temperatures increase by about 3.6°C by 2100 (SSP3-7.0). This scenario is marked by high population growth, limited regulations, and where countries do not collaborate on tackling climate change, resulting in elevated greenhouse gas emissions and a more severe impact on the climate.
  Mid term, 2041–60Long term, 2081–2100
SSPScenarioBest estimate (°C)Very likely range (°C)Best estimate (°C)Very likely range (°C)
SSP1-2.6Low GHG emissions


 		1.71.3 – 2.21.81.3 – 2.4
SSP2-4.5Intermediate GHG emissions2.01.6 – 2.52.72.1 – 3.5
SSP3-7.0High GHG emissions


 		2.11.7 – 2.63.62.8 – 4.6

Table 1 Shared socioeconomic pathways used in CSI

More information on these emissions scenarios and the transition from Representative Concentration Pathways (RCPs) to Shared Socioeconomic Pathways (SSPs) is available in this infographic for non-technical users and our dedicated factsheet.

How do the CSI projections differ from other climate projection data sets?

There are many different climate projection datasets available for different locations across Australia, and these may use different methodologies to suit different purposes. There are no right or wrong data sources, and each approach addresses different scientific uncertainties.  

The National Environmental Science Program’s Climate Systems Hub (NESP CSH) has produced a guide to help users understand the various modelling efforts being undertaken across Australia. This guide includes information on national datasets that cover Western Australia, and those produced by other modelling groups across Australia.  

Efforts to coordinate climate modelling across Australia are being made through the National Partnership for Climate Projections (NPCP). The NPCP aims to develop a consistent approach to deliver comparable, robust and fit-for-purpose future climate information to assess climate risks and inform adaptation planning. A factsheet with further information on the NPCP and some of these modelling efforts is available here. 

What projections do I use for water applications?

The Guide to future climate projections for water management in Western Australia (the guide, published by the Department of Water and Environmental Regulation) generally recommends using the National Hydrological Projections (NHP). The NHP use CMIP5 projections and have been validated in hydrological models. They offer projections for several water-specific variables (e.g. soil moisture) which may be useful inputs to an assessment. 

Advice in the guide will be reviewed as CSI projections, and others, become available, and detailed assessments of CMIP6 projections for hydrological modelling have been completed. 

Initial assessments suggest that CMIP5 and CMIP6 projections are similar for WA, so both can be applied to water assessments with confidence. 

The guide describes a framework for a systematic climate impact assessment, including how to choose and apply climate projections. The framework is applicable for any set of climate projections. 

How do I access the dataset available on SLIP/Data WA and what formats is the dataset available in?

The data is available at a 20km resolution for technical users to download from the Shared Location Information Platform (SLIP), the WA Government’s platform for sharing location-based information. The data is available in the following formats:
  • CSV (Excel)
  • Shapefile
  • NetCDF
  • Geodatabase

A technical fact sheet is also available on the SLIP to assist users to utilise the data.

How many years of data are available and what is the baseline?

Data is available for the historical period from 1951-2014. Future projections are available from 2015-2100.
 
A baseline period serves as a reference period from which the current or future climate is calculated. The baseline for CSI climate projections data is 1990-2009, consistent with NSW and Australian Regional Climate Modelling Project (NARCLiM) data covering NSW and south-eastern Australia.
 
A baseline period typically spans 20-30 years to understand changes to the climate while accounting for natural climate variability. It helps provide a consistent reference point to compare changes to over time.
 
It is important to also consider historical temperature changes prior to the baseline period. More information on these can be found in the 2024 State of the Climate report.
 
The baseline period is a starting point used to measure change. If you compare future temperatures to an earlier baseline (such as a cooler time in the past), the amount of warming may look larger. Earlier baselines tend to show larger increases in temperature.  

How are local climate projections developed and why do we need them?

CSI’s projections are being developed by scientists in the Department of Water and Environmental Regulation (the department) and Murdoch University’s Centre for Atmospheric Science. The projections are being produced in WA on one of the world’s most powerful supercomputers, the Pawsey Supercomputing Research Centre.

Global climate model information is processed in partnership with the NSW Department of Climate Change, Energy, the Environment and Water, via the NSW and Australian Regional Climate Modelling (NARCliM) program. This partnership provides the 20 km source data for further processing by the department and Murdoch University scientists to create the 4 km projections for WA.

The CSI follows the Coordinated Regional Climate Downscaling Experiment (CORDEX) design for downscaling to 4 km resolution. This is international best practice. For WA, the model is validated using historical datasets from weather stations across the state, as well as reputable weather forecasting models (i.e. ERA5 and WRF), to ensure outputs are as accurate as possible.

The coupled model intercomparison project (CMIP) provides a framework to analyse and validate GCM outputs. Its latest version, CMIP6 is used in the CSI and provides:

  • an improved representation of climate processes
  • higher resolution, enhanced simulation of extreme events
  • better uncertainty quantification
  • incorporation of Earth system components
  • integration of cutting-edge scientific advancements.

CMIP6 models represent a new generation of climate models and are highlighted in the most recent report from the Intergovernmental Panel on Climate Change (released in sections from 2021 to 2023).

For more information on:

What are the advantages of using dynamically downscaled models produced by the Climate Science Initiative?

Scientists need to use a process called ‘downscaling’ to look at regional climate influences and information such as topography at a local scale. Figure 5 explains the process of why downscaling is necessary and provides advantages to understanding future climate. This is one of the infographics produced by the CSI also available here.
 
Figure 4 Downscaling infographic showing how Global Climate Models are downscaled to create regional climate projections
 
The Climate Science Initiative uses dynamical downscaling to ensure the best understanding of WA’s future climate at a 20-km and 4-km resolution is available to decision makers.
 
This approach uses convection-permitting models (CPMs), a type of high resolution model which helps improve the resolution and accuracy of climate and weather projections by providing a more realistic representation of convection. These models are better able to simulate hourly precipitation characteristics that can be poorly represented in coarser-resolution climate models.
 
Because CPMs operate at a much finer scales, this allows them to resolve small-scale atmospheric processes like convective storms. The improved representation of convective storms also has implications for projections of wind, hail, fog, and lightning, and enable more accurate forecasts of intense weather events.
 
There are several critical areas where high-resolution models may be needed for providing reliable future assessments. These include transport, wind energy and other areas with considerations of flooding.
 
Another method for downscaling global climate model’s is statistical downscaling. This method involves using observed data and output from large-scale climate models to establish a relationship between the global and local climate. This means statistical downscaling methods rely on past data and assumptions and assumes these historical relationships will remain constant in the future, which may not be the case. Both statistical and dynamical downscaling methods are useful, depending on your purpose for using the information and variable of interest. However, it should be noted that if regridding is performed without new modelled information (including processes at convective scales), important climate related information may become unavailable, usually provided effectively by the convective permitting models. Further details can be found in the below peer-reviewed literature:

How should I use the climate projections effectively, and what are some common pitfalls to avoid?

There are a few considerations to be aware of to ensure effective use of climate projections data. There is no single textbook method to determine the best way to use all of the climate modelling data available, however, it is important to consider that there are diverse uses for climate projections in various fields such as water resources, the built environment and biophysical systems. Depending on the application, “fit for purpose” or effective use of information is different.
 
Once you have decided the purpose of your assessment and its application (including clear understanding of the climate change risks you want to evaluate), these are some general recommendations for how to use the climate data:  
 
  • For each regional climate model, we recommend using the full range of projections data available, i.e., all CMIP6 models downscaled, and all available future climate scenarios (SSPs). This will provide a solid evidence base to assess the impacts of climate change for risk assessments or other decision-making activities.  
  • It is not recommended to use the output from only 1 regional model, i.e., downscaling a single CMIP6 GCM for one particular future climate scenario. This can be misleading because climate projection data varies depending on the climate model and scenario selected. Different models and scenarios will produce slightly different outcomes, so a range is recommended to capture this uncertainty.
  • Uncertainties will always exist within the modelling due to climate variability and emissions uncertainty, so there are model-to-model differences. While climate projections provide a solid evidence base, they should be used as a guide to the future, and changes above or below the projected range should still be considered when managing risk.
  • We generally do not recommend using data from a singular grid point when using the climate projections data. Rather you should use a region of interest large enough around the point of interest. If data from a single grid point is used, it is very important to check if this grid point is over land or the ocean if the location is close to the coast. It is also critical to check the dominant land-use type at that grid point as it may not necessarily correspond exactly to the actual land-use at that point in the real world.
     
It is important to note that the CSI is providing the projections data. Users of the data such as government agencies and sectors will need to undertake further analysis and targeted impact studies for their purpose. For example, they may conduct risk assessments or seek to understand the implications of projected climate conditions on infrastructure and the broader economy, community, and environment.

Are there any limitations of CMIP6 models I should be aware of?

The new CMIP6 climate models provide incremental improvements in the representation of Australian temperature and rainfall patterns and their interactions, although some important model ‘biases’ and limitations remain as continuing scientific challenges to solve, including:
  • Some regional biases remain, for example excessive rainfall over the warm tropical seas in south east Asia (see the Maritime Continent) and rainfall pattern biases in the nearby tropical convergence zones (Grose et al., 2020).
  • Simulations of large-scale climate drivers and Australian rainfall could be further improved (Grose et al., 2020 and Di Virgilio et al., 2022).
  • Grid point values over water bodies are often not representative of land areas due to differences in physical properties and climate influences between land and water. Therefore, values from cells centered over major water bodies were excluded. Removing these values helps ensure that the data used for land-based climate modelling is more accurate and relevant.
     
The Climate Science Initiative has carefully chosen CMIP6 models that best represent the Australian climate, to reduce the impacts of these limitations.  Additionally, great care has been taken to select CMIP6 models that are statistically independent from each other. Please refer to Selecting CMIP5 GCMs for CORDEX Dynamical Downscaling: Model Performance, Independence, and Climate Change Signals (Di Virgilio et al., 2022) for further details.
 
In most cases, CMIP6 projections will be preferred as they represent the latest available climate projections information. CMIP6 projections broadly align with those from CMIP5. There are small improvements in confidence and clarity for some geographical areas. Using CMIP5 may be more appropriate if: reviewing or updating an existing risk assessment based on CMIP5 projections; for compliance with policy specifying RCPs should be used; or where application requires the consideration of a very high emissions scenario and use of RCP8.5 may be useful.
 
Care should be taken when using both CMIP5 and CMIP6 in the same analysis, because there are differences between baselines and future scenarios which can make interpretation difficult.

What is the difference between bias adjusted and non-bias adjusted data?

Models may produce results where some regions are slightly hotter (cooler) or drier (wetter) when compared to observations of the climate, and these differences are known as ‘model bias.’ Climate scientists conduct bias adjustment by comparing climate model data for a historical time period with observed actual climate records and then adjusting the model so that it more closely matches the observations. However, bias adjustment cannot correct all sources of bias, and is most useful when large systematic biases are present. In some instances when biases are small, applying bias adjustment can sometimes introduce or amplify bias.  

Depending on the bias adjustment method used, it can also alter the change in climate as compared to the original model output (raw data). For instance, some methods adjust individual variables or single statistical distributions, which can affect the physical consistency of the dataset because it altered the relationships between variables over time. This is an important consideration when selecting and using bias-adjusted data.  

Bias adjustment is also limited to variables with good observational data coverage, such as temperature and rainfall. It is not possible to bias-adjust all model variables, including where observations are unavailable or limited (for example, soil moisture or subsurface temperature).

The current monthly 20 km data is bias adjusted for temperature and rainfall. 

How do I access the new dataset available on National Computational Infrastructure (NCI)?

The 4 km South West Western Australian (SWWA) dataset is publicly available via the NCI THREDDS server: NCI Data Catalogue - CSI 

The data is provided in netCDF format and is intended for technical users who wish to analyse it using programming languages such as R, Python, Matlab or similar. Examples of how to analyse the data using Python workflows (using Jupyter notebooks, which are open source and free to use) are available on this GitHub page.   

Data can be analysed directly on the THREDDS server without downloading files. However, for computationally intensive workflows requiring significant memory or processing time, users may need to either: 

  • download the data to a local system with sufficient computational capacity, or 
  • run analyses on the National Computation Infrastructure (NCI), if they have an active project and Service Unit (SU) allocation. 

A smaller subset of the data is also available on an Australian Research Data Commons (ARDC) Nectar cloud server (32 CPUs, 126 GB RAM, 10 TB storage). This option is suitable for users experienced with netCDF data. 

To request access to the ARDC Nectar environment, contact: climatescience@dwer.wa.gov.au. 

What is the difference between the 4 km data for South West WA and the 20 km statewide CSI projections?

Both datasets provide valuable climate information but differ in spatial resolution and geographic coverage, making them suitable for different applications. 

The 20 km statewide projections cover the entire Western Australian domain, offering a consistent, state-wide view of future temperature and rainfall. This makes them well suited to broad applications such as statewide (including sectoral) risk assessments, policy development, and strategic planning where full spatial coverage is required. 

In contrast, the 4 km South West WA (SWWA) projections provide much finer spatial detail. Their higher resolution allows for improved representation of local features, such as topography, resulting in more detailed and realistic simulations. Consequently, the 4 km dataset is better suited to local-scale analysis, planning, and risk assessments within the South West (see Figure 3 for the SWWA 4 km coverage area). 

As the 4 km North West WA (NWWA) datasets become available, they can be incorporated into analyses where relevant. 

Each dataset represents a plausible future climate using different modelling approaches. As with all climate models, both contain biases relative to observations, and some variables (such as precipitation) may differ between the two datasets. These differences reflect the inherent uncertainty in climate projections and highlight the value of using multiple datasets where appropriate. 

How were the SWWA and NWWA domains and boundaries developed?

The South West Western Australia (SWWA) and North West Western Australia (NWWA) modelling boundaries define where high‑resolution (4 km) climate projections are produced under the Climate Science Initiative (CSI) (see Figure 3).

Figure 4 Climate Science Initiative modelling domains

Figure 4 Climate Science Initiative modelling domains 

The CSI uses a nested regional climate modelling approach consistent with the New South Wales (NSW) and Australian Regional Climate Modelling Project (NARCliM) project and international best practice under the Coordinated Regional Downscaling Experiment (CORDEX). In this approach, global climate model and reanalysis data are first downscaled to a 20 km regional domain over Australasia, through a CSI and NARCliM2.0 partnership. These outputs then provide boundary conditions for further downscaling to 4 km resolution in targeted subdomains for NSW and WA. 

The SWWA and NWWA domains were selected through stakeholder consultation, user needs assessment, and consideration of high‑performance computing constraints. Simulating the entire state at 4 km resolution was not feasible, so two subdomains were defined. Together, the two domains cover almost one third of Australia and allow computational resources to be focused where higher-resolution information adds the most value.  

The SWWA domain was prioritised for delivery due to strong stakeholder demand and feasibility, with the NWWA domain following as the next phase. 

Within the broader 20 km domain, the SWWA and NWWA regions were selected as convection‑permitting subdomains to: 

  • capture key climate regimes and gradients across WA, 
  • focus on areas with high population, infrastructure, ecological, and water-resource sensitivity, and 
  • support scientifically robust high-resolution modelling. 

This two-stage downscaling approach enables detailed, locally relevant climate projections while maintaining consistency with national and international modelling frameworks. 

What climate variables will the CSI simulations produce, and how will these data be made available?

The Climate Science Initiative (CSI) uses the Weather Research and Forecasting (WRF) model to dynamically downscale climate projections from Global Climate Models (GCMs) that contributed to the Coupled Model Intercomparison Project phase 6 (CMIP6). The WRF simulations realistically represent key climate processes and generate over 150 climate variables. A full list of these variables is available here. The most frequently requested climate variables (CORDEX core variables) including rainfall, temperature, humidity, evaporation, and wind, are available on NCI. 

Guidance materials for technical users including technical and user guides are now available at https://github.com/DWER-WA/CSI-Overview.git. This overview details the model configuration, data access, data organisation, list of variables, citation and license. These materials will be updated periodically. 

These pages will be updated on a regular basis noting some guides are still in preparation.

Reference list

Di Virgilio, G., Ji, F., Tam, E., Nishant, N., Evans, J. P., Thomas, C., et al. (2022). Selecting CMIP6 GCMs for CORDEX dynamical downscaling: Model performance, independence, and climate change signals. Earth's Future, 10, e2021EF002625. https://doi.org/10.1029/2021EF002625
 
Grose, M.R., Narsey, S., Delage, F.P., Dowdy, A.J., Bador, M., Boschat, G., Chung, C., Kajtar, J.B., Rauniyar, S., Freund, M.B., Lyu, K., Rashid, H., Zhang, X., Wales, S., Trenham, C., Holbrook, N.J., Cowan, T., Alexander, L., Arblaster, J.M. and Power, S. (2020) 'Insights from CMIP6 for Australia's future climate', Earth's Future, 8(5), e2019EF001469. doi: 10.1029/2019EF001469.
 
Kendon, E.J., Ban, N., Roberts, N.M., Fowler, H.J., Roberts, M.J., Chan, S.C., Evans, J.P., Fosser, G. and Wilkinson, J.M. (2017). Do Convection-Permitting Regional Climate Models Improve Projections of Future Precipitation Change? Bulletin of the American Meteorological Society, 98(1), pp.79–93. https://doi.org/10.1175/bams-d-15-0004.1.
 
United Nations Environment Programme (UNEP) (2022) 'Emissions Gap Report 2022', United Nations Environment Programme. Available at: https://www.unep.org/resources/emissions-gap-report-2022 (Accessed: 17 December 2024).

Have a question or want to report a problem?

Fill in the form to get assistance or tell us about a problem with this information or service.

Send feedback