Key points
- NARCliM creates world-class regional climate projections by using the latest available climate models adopted by the Intergovernmental Panel on Climate Change (IPCC).
- NARCliM combines carefully selected global and regional climate models to simulate a range of plausible future climates.
- The rigorous selection process balances the need for a suitable range of simulated climates with the time and resources needed to compute and release the vast amounts of data.
Climate models
Climate models are a widely adopted tool to investigate past and future climates to help focus climate change adaptation and mitigation efforts. Climate models have existed for more than 60 years and have evolved over time as our understanding of the climate system improves and as technology advances.
NARCliM uses a selection of global climate models (GCMs) and regional climate models (RCMs) to build its regional climate projections. Together, GCMs and RCMs provide the broad-scale climate patterns and local information that are needed to capture the climate of NSW and the broader Australasian region. To create locally relevant data, GCMs and RCMs are combined through a downscaling process.
Each NARCliM generation relies on the latest available GCMs and RCMs covered by the World Climate Research Programme (WCRP) framework. Under the framework, the development of GCMs is covered by the Coupled Model Intercomparison Project (CMIP; which are adopted by the IPCC), and RCMs by the Coordinated Regional Climate Downscaling Experiment (CORDEX). CMIP and its associated data infrastructure have become essential to the Intergovernmental Panel on Climate Change (IPCC) and other international and national climate assessments.
GCMs used by NARCliM
Each NARCliM generation is built using a carefully selected subset of GCMs from the latest available CMIP phase. Using a subset of GCMs means that the NARCliM modelling process balances the need for a range of future scenarios with computing time and resources.
GCMs are selected based on 3 factors. They must:
- accurately represent the climate of our region
- be statistically independent from one another
- accurately represent the climate diversity for the Australian context.
GCM selection is backed by peer review – for example, the robust GCM selection process for NARCliM2.0 was published in a leading high-impact journal, Earth’s Future.
NARCliM uses the following CMIP GCMs:
- CMIP6 – NARCliM2.0 uses 5 GCMs from CMIP6 (ACCESS-ESM1-5, EC-Earth3-Veg, NorESM2-MM, MPI-ESM1-2-HR, UK-ESM1-0-LL).
- CMIP5 – NARCliM1.5 uses 3 GCMs from CMIP5 (ACCESS1.0, ACCESS1.3, CanESM2). These were selected following a similar evaluation and ranking process, and to complement NARCliM1.0 by adding to the spread of plausible future climates. This means that NARCliM1.5 adds to the spread of projections first presented in NARCliM1.0.
- CMIP3 – NARCliM1.0 uses 4 GCMs from CMIP3 (MIROC3.2, ECHAM5, CGCM3.1, CSIRO-Mk3.0). These were selected through a peer-reviewed evaluation and ranking process, to identify the GCMs that most closely matched observed climate data across NSW.
GCM selection is governed by a 3-step process.
Selecting GCMs for NARCliM
For each NARCliM generation, only the GCMs that have the required data outputs for all emissions scenarios of interest are initially considered. For example, NARCliM2.0 considered CMIP6 GCMs with data outputs for SSP1-2.6 and SSP3-7.0.
Dozens of GCMs are assessed for consideration in each NARCliM generation. Every GCM has its biases (instances where the model data doesn’t align with observed past climate data). Selecting multiple GCMs helps to address these biases – where one model has a bias for a specific variable, another model may perform better. This results in a more suitable range of plausible future climates.
Many GCMs are built from a shared computer code base and address the simulation of physics (and their uncertainties) in a similar manner. This means that different GCMs may simulate very similar future temperature and precipitation over Australia. The more similar the models are to one another, the less independent they are. Greater independence is needed in the subset to reflect the best spread of plausible climate futures.
After the appropriate GCMs are identified, RCMs are chosen based on their capacity to simulate the south-eastern Australian regional climate, to dynamically downscale the selected GCMs.
RCMs used by NARCliM
To provide locally relevant information, NARCliM applies the 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.
NARCliM uses the following WRF versions:
- NARCliM2.0: 2 RCMs (WRF4.1.2)
- NARCliM1.5: 2 RCMs (WRF3.6.0.5)
- NARCliM1.0: 3 RCMs (WRF3.3)
Selecting RCMs for NARCliM
As with GCMs, only a subset of RCMs is used to reduce computational time and energy, while supporting NARCliM’s production of a valuable selection of plausible future climates.
RCMs are selected by considering:
- the performance of each RCM (how well it simulates the south-east Australian climate)
- the spread of distinct climates they contribute to (to increase the range of plausible future climates using the fewest RCMs possible).
For NARCliM2.0, 78 RCM configurations were tested. Testing included 3 phases, where each test phase was informed by the outcomes of the previous test. This process systematically removed poor-performing RCM options to leave only the most appropriate combinations.
More information
Further detail is in the following reports:
- Peer-reviewed literature on the GCM selection process for NARCliM2.0 is published in Earth’s Future.
- Peer-reviewed literature on the RCM selection process for NARCliM2.0 is published in Geoscientific Model Development.
- Technical information on GCM and RCM selection for NARCliM1.5 can be found in the NARCliM1.5 Technical Methods Report.
- Technical information on the GCMs for NARCliM1.0 can be found in the NARCliM Technical Note 1.
- Peer-reviewed literature on the RCM selection process for NARCliM1.0 and 1.5 is published in Geoscientific Model Development Discussions.