Introduction to Application-ready Data

Data that relate to a particular location and have a format similar to historical data (including natural variability)

Certain applications need more than general information about projected changes in the climate. These applications require projections as input to further analyses or modelling to explore the impact of particular scenarios in detail. This requires application-ready, locally-relevant future climate data:

1. Application-ready: data in a form compatible with the applied model or analysis, including a representation of climate variability compatible with the data used to calibrate the model
2. Locally-relevant: made relevant to the local spatial scale of interest, including the local climate influences (e.g. accounting for local topography)

The choice to use application-ready data and the method to produce it are part of a wider set of choices in using climate projections . In making these choices, it is important to have an overall clear perspective on projections and the sources of uncertainty and confidence . The Decision Tree can help you make the right choice and guide you to the right pages on this site. If detailed datasets are needed, the Regional projections data pathway can be used.

Global climate model (GCM) outputs can’t be used directly in applied analyses that require calibration to local observations. This is because GCMs have a coarse spatial scale and contain some difference to observations, or ‘bias’. Some bias is expected and does not mean that the projected changes are unreliable (for more detail, see Modelling and projections ).

There are a range of methods for producing appropriate datasets from GCM outputs, in two main categories, as described in the table below.

SCALED DATA AVAILABLE IN CCIA

Application-ready data created from both the current (Coupled Model Intercomparison Project phase 6, or ‘CMIP6’) and previous (Coupled Model Intercomparison Project phase 5, or ‘CMIP5’) generations of GCMs are available from Climate Change in Australia. The data creation approaches and the data availability differ slightly between CMIP5 and CMIP6 application ready datasets (see below and on the Data Availability page).

Either mean scaling or quantile-quantile scaling was applied to 30-year observed time-series datasets to produce time-series data for future periods, with the scaling method used depending on the generation of GCMs (CMIP6 or CMIP5) and the climate variable. No transient change has been used, meaning that the future time-series data can be regarded as representative of the mean state of the future climate. This makes these datasets ideal for deriving daily statistics for the future climate. For example, the Thresholds Calculator uses the CMIP5 application-ready datasets to derive data on the average number of days per year above or below particular temperature and rainfall thresholds.

The best way to use application-ready data depends on the field of research, the relevant aspects of climate change relevant to the research question and the type of applied model that is used. Examples of choosing an overall approach, including the method of producing application-ready datasets may help. We have produced three case studies covering a range of sectors:

Case study: Hydrology

Case study: Biodiversity

Case study: Human health (heat)

It is important to note that the fine spatial and temporal details in application-ready data are derived from the observed datasets, not the climate models. For example, Bureau of Meteorology observed temperature and rainfall data are available on a 5km (approx.) grid, while the projected changes in temperature and rainfall from the CMIP6 and CMIP5 GCMs used have resolutions of at least 80km. When combining the observed and model data, the model data are first interpolated to a 5km grid (which does not alter the climate change patterns) then applied to the observed data. Hence, the application-ready data are simply modified observed data. A detailed description of the scaling methods used is provided here.

CMIP6 application-ready data

Quantile-quantile scaling was applied to 30-year observed time-series datasets, centred on 2000 (1985–2014) to produce time-series data for future periods centred on 2050 (2035-2064) and 2085 (2070-2099). In all cases, the changes applied were calculated using the baseline period 1985-2014.

The Australian Climate Service has funded the creation of data for a selection of key climate variables for which appropriate observed climate data are available; and are of sufficient quality and coverage. The observed data are sourced from two Bureau of Meteorology gridded datasets: AGCD, a ~5km-resolution observed dataset, and BARRA-R2, a ~12km-resolution reanalysis dataset. Both datasets are used for temperature and rainfall. The ~5km-resolution AGCD-based version of the application-ready data incorporates the highest resolution observed data for temperature and rainfall. It is suitable for applications that require information on how the climate varies on scales less than ~12km for these variables. The BARRA-R2-based version provides data that is consistent with other variables for which AGCD data are not available (i.e., relative humidity, solar radiation, wind speed). For convenience, the BARRA-R2-based version has been interpolated (using bi-linear interpolation) to the same ~5km-resolution grid of the AGCD-based version. However, the interpolated dataset still only contains information on how the climate varies on scales of ~12km or more. If a user already uses AGCD or BARRA-R2 for historical analysis, they may wish to choose the version of the application-ready future data that matches this to ensure consistency between analyses for past and future time periods.

1. Australian Gridded Climate Data (AGCD) 0.05° gridded data (Evans A, Jones DA, Smalley R, Lellyett S, 2020 'An enhanced gridded rainfall analysis scheme for Australia’. Bureau of Meteorology Research Report No. 041)
2. BARRA-R2 reanalysis 0.11° gridded data (Su C-H, Rennie S, Dharssi I, Torrance J, Smith A, Le T, Steinle P, Stassen C, Warren RA, Wang C, Le Marshall J, 2022 ‘BARRA2: Development of the next-generation Australian regional atmospheric reanalysis’. Bureau of Meteorology Research Report No. 067), interpolated to 0.05° using a bi-linear method

The CMIP6 application-ready data use simulated climate changes from a subset of nine CMIP6 climate models selected for dynamical downscaling over Australia by the Australian Climate Service and the NSW Government’s NARCliM2.0 project. They perform relatively well at simulating relevant aspects of the climate and represent some of the potential range of changes in annual mean temperature and rainfall over Australia (Grose et al 2023). This subset reduces the effort required for data management, while still sampling some of the range of uncertainty from the full set of ~50 CMIP6 GCMs.

CMIP5 application-ready data

Depending on the variable, mean scaling or quantile-quantile scaling was applied to 30-year observed time-series datasets, centred on 1995 (1981–2010) to produce time-series data for the future periods 2016-2045, 2036-2065, 2056-2085 and 2075-2104. In all cases, the changes applied were calculated using the baseline period 1986-2005.

Data are provided for all variables for which appropriate observed climate data were available to scale; and were of sufficient quality and duration. The observed data come from a variety of sources, sometimes in gridded format and sometimes for sites (see the table below and the Data Delivery brochure ).

3. Australian Water Availability Project (AWAP) 0.05° gridded data (Jones DA, Wang W, Fawcett R 2009 'High-quality spatial climate data-sets for Australia.' Australian Meteorological and Oceanographic Journal 58, 233-248.)
4. Bureau of Meteorology (BoM) high-quality station network (for details, see the Data Delivery brochure , p12)
5. ERA-Interim reanalysis 0.75° gridded data, interpolated to 0.05° using a bi-linear method
6. CSIRO Land and Water 0.05° gridded data (Teng J, Vaze J, Chiew FH, Wang B, Perraud J-M (2012) 'Estimating the relative uncertainties sourced from GCMs and hydrological models in modeling climate change impact on runoff.' Journal of Hydrometeorology 13, 122-139.)

The CMIP5 application-ready data can be made available for averages and time-series over a range of spatial scales. Spatial detail ranges from Cluster-average, to a 5 km grid-average, to specific cities and towns (limited to sites with high-quality baseline data). Changes are based on a subset of eight CMIP5 climate models that simulate most of the range of changes in seasonal-mean temperature and rainfall over most of Australia (Technical Report Chapter 5 Box 9.2), plus downscaling where appropriate. This subset reduces the effort required for data management, while still sampling most of the range of uncertainty from the full set of 40 climate models.