eReefs Marine Model Results
The results datasets from runs of the GBR4 and GBR1 eReefs marine models are published as National Reference Datasets via the NCI.
Both models use the CSIRO Environmental Modelling Suite (EMS) software to simulate the waters of the Great Barrier Reef. They are first run to simulate the hydrodynamics of the reef, producing a results dataset containing ocean current data, as well as temperature, salinity and sea-surface temperature.
The hydrodynamic model results are then used as boundary forcing for transport models, which can simulate river tracers for the various rivers which flow into the Great Barrier Reef lagoon or calculate the biogeochemical and optical properties of the water column and sediments.
Model results are published as one dataset per unique combination of model grid (GBR4 or GBR1), forcing data selection and run mode. Dataset names follow our Model simulation naming protocol which embeds some semantic information about that configuration in the dataset ID.
The raw results files are published as netCDF files that comply with the NetCDF Climate and Forecast (CF) Metadata Conventions. Each file represents either a single day or a single month of the simulation, and includes the simulated date in the filename.
All the datasets are licensed as Creative Commons by Attribution (CC-BY 4.0), and may be freely referenced by other data products. Please refer to the metadata record for each dataset for the preferred citation.
If you use any of these eReefs marine model results datasets in your research, please include the following acknowledgement:
The eReefs datasets (model simulations, satellite or in-situ observations) and software were produced as part of the eReefs project (ereefs.org.au), which is a collaboration between Australia’s national science agency CSIRO, the Australian Institute of Marine Science (AIMS) and the Queensland Government, with observations obtained through the Integrated Marine Observing System (IMOS) and the Great Barrier Reef Marine Park Authority (GBRMPA). eReefs is funded by the Australian Government’s Reef Trust.
GBR4 Regional Model Datasets
GBR4 Regional Model Grid
All GBR4 Regional Models are based on a curvilinear grid with a spatial resolution of approximately 4km. This grid is made up of 220 x 500 horizontal cells and 47 depth layers with 1m depth resolution at the ocean surface. The grid extends into the coral sea to ensure that simulations based on it include the south equatorial current while avoiding continental shelf features. The GBR4 grid definition is available as a geospatial layer which you can access via links in its metadata record here.
Current GBR4 Regional Model Results
The best available eReefs Marine Hydrodynamics dataset is a hindcast based on version 4.0 of the GBR4 Hydrodynamic model (GBR4_H4p0).
This version of the model includes several enhancements over the v2.0 model, including updated OBC parameterisation allowing free-developing realistic shelf-break currents (no relaxation), and updated heat penetration via use of the Sed-BGC-optic model to calculate a spatially and temporally variable short wave radiation attenuation parameter. These hindcasts are derived from validated boundary forcing data which is the best available for the time period of the simulations, including:
- GBR100 DeepReef Bathymetry
- Ocean boundary forcing from the BRAN 2020 Reanalysis results of the Bluelink global ocean model
- Atmospheric forcing from the BARRA R2 global atmosphere model
- Tidal forcing from the global TPXO tide model
- River Flows and nutrient loads from a combination of the eReefs Grid2Grid hydrology model and the Paddock to Reef SOURCE catchments model
The GBR4_H4p0 hindcast datasets will be extended from time to time as new validated forcing data is published.
All current GBR4 Regional Model Results datasets can all be visualised and queried via the eReefs Data Explorer.
| Dataset (metadata link) | Start | End | Temporal Resolution |
Size (TB) | Links |
|---|---|---|---|---|---|
| NEW! Released May 2025 GBR4 Hydrodynamics v4.0 Hindcast ( GBR4_H4p0_ABARRAr2_OBRAN2020_FG2Gv3_Dhnd) |
2010-09 | 2022-11 | hourly | 2.6 TB | Data Explorer NcML Aggregation NetCDF Files |
| NEW! Released December 2025 🡒 GBR4 Biogeochemistry and Sediments v4.2 Baseline Catchment Scenario ( GBR4_H4p0_ABARRAr2_OBRAN2020_FG2Gv3_B4p2_Cq5b_Dhnd) |
2011-01 | 2022-10 | daily | 1.1 TB | Data Explorer NcML Aggregation NetCDF Files |
| Coming soon! 🡒 GBR4 Hydrodynamics v4.0 River Tracers ( GBR4_H4p0_ABARRAr2_OBRAN2020_FG2Gv3_Dhnd_Rivers) |
2010-12 | 2022-10 | daily |
Deprecated GBR4 Near-Real-Time Model Results
Version 2.0 of the GBR4 hydrodynamic model was run in near-real-time mode from 2015 to January 2024, appending new files to the results datasets every day that new forcing data became available.
While the model was operating, the time lag between ‘now’ and the most recent timestamp available in the dataset varied depending on how long it took us to acquire and ingest the third-party forcing data, but was usually about one week, with the BGC model a day behind the hydrodynamics.
Unfortunately, flooding events in northern Queensland catchments during the 2023/24 summer monsoon season caused damage to the real-time streamflow and water quality monitoring network in the Great Barrier Reef catchments, which meant we no longer had access to one of our most critical forcing datasets. We were initially able to restart the model with a simulated dataset substituting for the data lost to TC Jasper in December 2023, but were then faced with the failure of the Normanby river gauge on January 18 2024 and the Daintree gauge on February 8 2024. Without flow data for these major rivers, the hydrodynamic model results became unreliable, and so we reluctantly made the call to cease operation of the GBR4 near-real-time models, and complete the datasets with the end date of January 17 2024.
The GBR4_H2p0 datasets have been superseded by the suite of current GBR4 datasets (listed above) which use much higher quality, validated forcing data and an improved mathematical model.
These datasets will be removed from the NCI fx3 project and THREDDS server in June 2026.
After that date, the metadata records for these datasets will remain online and available for reference purposes, but the data files will no longer be easily accessible. We recommend that any researchers who currently depend on these datasets switch to using the newer GBR4_H4p0 datasets as soon as possible.
Deprecated GBR4 Water Quality Scenario Results
These datasets were produced using a succession of versions of the GBR4 Biogeochemistry and Sediments model to support multiple editions of the Reef Water Quality Report card.
Each related set of scenarios used the accumulated results of the then-current near-real-time hydrodynamic model as forcing data, but calculated in hindcast mode using the best-available validated (for the baseline) or scenario catchment forcing data and nutrient loads.
Each new suite of scenario results completely supersedes the one before it, and all of these have now been superseded by the new GBR4_H4p0_B4p2 datases listed above.
We recommend that you do not use these for new research projects. The metadata records for these datasets will remain online and available for reference purposes, but the data files are removed from the NCI fx3 project and THREDDS server once the succeeding datasets have been published. If you need access to these old datasets for research purposes, please use the contact form on this website to request access to CSIRO’s archive copies.
The GBR4_H2p0_B3p1 catchment scenario datasets will be removed from the NCI fx3 project and THREDDS server in June 2026. If you are still using these for your research, please switch to the GBR4_H4p0_B4p2 catchment scenario results as soon as possible.
GBR1 Shelf Model Datasets
The GBR1 Shelf Model is a higher-resolution model than GBR4, and uses results from the GBR4 models as ocean boundary forcing data.
GBR1 Shelf Model Grid
The GBR1 model grid is a curvilinear grid with a 1km spatial resolution made up of 510 x 2390 horizontal cells and 44 depths, with a 1m vertical resolution at the surface. The grid extends far enough from the Queensland coast to allow the model to resolve continental shelf hydrodynamics and larger reefs and islands, but does not include the rest of the Coral Sea.
Although only 50% of surface cells and 22% of all cells are wet, GBR1 model runs still calculate values for about twelve million cells every timestamp, so are both quite slow to run and produce very large results datasets!
The GBR1 grid definition is available as a geospatial dataset which you can access via links in its metadata record.
Current GBR1 Shelf Model Results
Version 2.0 of the GBR1 shelf model was a part of the eReefs Near Real Time modelling workflow from 2015 onwards, with the model results datasets appended to each day as soon as the GBR4 Hydrodynamics v2.0 forcing dataset had been updated. While this system was operating, the GBR1 near-real-time model results usually lagged their GBR4 equivalents by only a few hours. When the GBR4 near-real-time modelling system had to cease operating on January 17, 2024, the GBR1 system lost access to its critical GBR4 forcing, and it also had to end.
The completed GBR1_H2p0 datasets are still the most recent GBR1 datasets available. They will remain online until the eReefs team complete work on the suite of GBR1_H4p0 datasets which will supersede them.
| Dataset (metadata link) | Start | End | Temporal Resolution | Size (TB) | Links |
|---|---|---|---|---|---|
GBR1 Hydrodynamics v2.0 NRT (GBR1_H2p0) |
2014-12-01 | 2024-01-17 | hourly | 20 TB | NcML Aggregation NetCDF Files |
🡒 GBR1 Hydrodynamics v2.0 NRT River Tracers (GBR1_H2p0_rivers) |
2014-12-01 | 2024-01-17 | daily | 3.8 TB | NcML Aggregation NetCDF Files |
| NEW! Released May 2025 🡒 GBR1 Diuron Dispersal Scenario ( GBR1_H2p0_Cq3pe1_Dhnd) |
2016-01 | 2018-06 | hourly | 34 GB | NcML Aggregation NetCDF Files More information about this pesticide scenario: Pesticide Modelling and Management |
Deprecated GBR1 Shelf Model Results
The following sets of GBR1 shelf model results are older and have been retired by the eReefs modelling team in favour of newer data products.
We recommend that you do not use these for new research projects.
The metadata records for these datasets will remain online and available for reference purposes, but the data files have been removed from the NCI fx3 project and THREDDS server, and the datasets are not available for visualisation in the eReefs Data Explorer. If you need access to these old data files for research purposes, please use the contact form on this website to request access to CSIRO’s archive copy of the dataset.
| Dataset (metadata link) | Start | End | Temporal Resolution | Size (TB) | Links |
|---|---|---|---|---|---|
GBR1 Biogeochemistry and Sediments v3.2 NRT (GBR1_H2p0_B3p2_Cfur_Dnrt)WITHDRAWN February 2020 due to abnormally high/unrealistic Chlorophyll-a levels on the shelf edge and reef matrix |
2019-10-16 | 2024-01-16 | daily | 7.2 TB | Best alternative is GBR4_H4p0_ABARRAr2_OBRAN2020_FG2Gv3_B4p2_Cq5b_Dhnd |
| GBR1 Biogeochemistry and Sediments v1.0 (v924) NRT | 2014-12 | 2019-11 | daily | 5.6 TB | Best alternative is GBR4_H4p0_ABARRAr2_OBRAN2020_FG2Gv3_B4p2_Cq5b_Dhnd |
GBR1 Hydrodynamics v1.71 (GBR1_H1p71) |
2014-12 | 2016-04 | hourly | 2.2 TB | Superseded by GBR1_H2p0 |
Accessing GBR4 and GBR1 Model Results
NCI THREDDS Catalog and Query Endpoints
We recommend that researchers access these datasets via the NCI’s THREDDS Server, which supports a number of standard HTTP data-query protocols, including OPenDAP, Web Map Service with netCDF Extensions, and Web Coverage Service (WCS). It is also possible to download the individual results files from this service, but we do not recommend that approach, as some of these datasets are very large (10s of TB). It is much more efficient to query just the subset of data you need!
For each dataset, the THREDDS catalog exposes both the individual NetCDF files and a time-based aggregation that allows the entire dataset to be queried from a single URL. See the tables below for links.
NCI High Performance Compute
Researchers with existing high-performance compute allocations on the NCI may apply for access to the fx3 project if you wish to use the eReefs model results dataset files in your HPC workflows on Gadi or from NCI Nirin Cloud servers. Please refer to the NCI User Guides for detailed instructions.
eReefs Tools and Tutorials
The eReefs platform includes a number of tools designed to help you access and use these datasets:
- The GBR4 and GBR1 model grids and current results datasets can be browsed, displayed, queried and animated via the eReefs Data Explorer.
- The emsarray python library understands the curvilinear grids used for GBR4 and GBR1 model results and can be used to query the datasets from the NCI THREDDS server without needing to download them. Detailed examples of how to use this librray to subset, plot and animate eReefs marine model results are available as Jupyter notebooks.
- The eReefs data extraction tool can be used to download environmental conditions at research sites on the Great Barrier Reef from the GBR4 and GBR1 model results.
- A growing number of tutorials explain how to work with this data in both Python and R.
Variables in GBR4 and GBR1 Datasets
Grid Dimensions and Coordinates
The GBR4 and GBR1 model results are all published as NetCDF files which comply with the Climate and Forecasting Metadata Conventions.
The published datasets all use the EMS Simple Results Geometry Convention, which uses the following NetCDF dimensions:
| Dimension Variable | Description |
|---|---|
i |
x-axis for the grid. This is an index for where a grid cell is positioned outward from the Queensland coast. It starts at 0 for grid cells furthest inland and gets higher as you move offshore. |
j |
y-axis for the grid. This is also an index which indicates where a grid cell is positioned along the Queensland coast. It starts at 0 for the most southward cells in the grid, and gets higher as you move northward towards Cape York. |
k |
z-axis for the grid cells that occur in the water column. This starts at 0 at the seabed, and gets higher as you move toward the sea-surface. |
k_sed |
z-axis for the grid cells that occur in the sediment layers on the sea floor. This starts at 0 at the bedrock, and gets higher as you move towards the sea-bed (i.e. k=0 for the water column) |
t |
This is the time index. It starts at 0 for the earliest timestep in the dataset, and increases from there. |
Each dataset will always contain a matching set of coordinate variables, which can be used to translate the NetCDF dimension indices into real-world coordinate systems:
| Coordinate Variable | Related Dimension(s) | Description |
|---|---|---|
longitude |
i,j |
The longitude in decimal degrees-east of the centre of the grid cell in the (i, j) position. |
latitude |
i,j |
The latitude in decimal degrees-north of the centre of the grid cell in the (i, j) position. |
botz |
i,j |
The depth of the sea-bed below the mean astronomical tide in metres at the horizontal grid cells in the (i, j) position. |
zc |
k |
The elevation in metres of the centre of any water-column grid cell with z-axis index k relative to the mean astronomical tide line (a.k.a. sea level). Positive values are above sea level, and negative values (almost all) are below sea level. |
zc_sed |
k_sed |
The elevation in metres of the centre of a sediment grid-cell at z-axis index k_sed relative to the location of the sea-floor, which has zc_sed=0. Negative values indicate distance below the sea-floor. |
time |
t |
The time that timestamp t represents in the simulation, given as days since 1990-01-01 00:00:00 +10 |
When you are working with these datasets, you will need to ensure your code properly translates the grid-cell-index dimensions (i, j, k, t) or (i, j, k_sed, t) which are used to extract the data-values to real-world coordinates (longitude, latitude, depth, time) before visualising the data or combining the eReefs model results with other datasets.
The emsarray python library has been created especially to help with this translation step: it supports working with data both grid and world coordinates, and greatly simplifies working with eReefs regional model results datasets.
BGC Data Variable Definitions
All other variables which are present in the eReefs Marine Model Results files are data variables which contain the results of the model calculations.
The Biogeochemistry and Sediments model results datasets may include several hundred data variables! For definitions of what each of these variables represents, you can refer to:
- The Biogeochemical Scientific Description for version
B3p0of the model. This was published at Geoscientific Model Development which contains an assessment of an eReefs configuration. - For additional details, especially definitions of diagnostic variables and simulated observations (true colour, Secchi depth etc.) as well as processes added to the model since 1 January 2020, please see EMSmanual_afterGMDarticle8Jan25.
- The
puv__parameterattribute from the netCDF metadata for each variable. This attribute is present in recent datasets (GBR4_H4p0and later), and takes the form of a URL for a formal vocabulary record, e.g. http://vocab.nerc.ac.uk/collection/P01/current/ARGTMOD1/. You can open these links in a browser to see a description and other information about the variable, and they also support content negotiation for machine-readable formats like JSON-LD, so can be used to look up variable level metadata from other software. Please see the NERC Vocabulary Server documentation for more information about these vocabulary terms.
A note on sediment variables:
Many BGC dataset variables are calculated for the both the water column and for sediment.
- The data variable for cells in the water column will have
(i, j, k, t)dimensions (zcas a depth coordinate) - The data variable for cells in sediment layers will have
_sedas a suffix on the variable name and(i, j, k_sed, t)as dimensions (zc_sedas a depth coordinate)
Key Data Variables by Field of Research
If you are planning on using the eReefs GBR4 or GBR1 datasets in your own research, these lists should help you identify the most important data variables for your field.
Water Quality
Chl_a_sum) from the GBR4_H2p0_B3p1_Cq3b_Dhnd dataset| Concept | Data Variable | Units |
|---|---|---|
| Total Suspended Solids | tss, tss_sed |
|
| Total Chlorophyll | Chl_a_sum |
|
| Nitrate | NO3, NO3_sed |
|
| Secchi from 488nm | Secchi |
|
| Temperature | temp |
|
| Salinity | salt |
- Load water quality variables in the eReefs Data Explorer
- Learn more about water quality measures in eReefs
- Learn how to plot eReefs data on a map and plot an eReefs transect
Carbon Chemistry
DIC) from the GBR4_H2p0_B3p1_Cq3b_Dhnd dataset| Concept | Data Variable | Units |
|---|---|---|
| Dissolved Inorganic Carbon | DIC |
|
| Total Alkalinity | alk |
|
| PH | PH |
|
| Aragonite Saturation State | omega_ar |
|
| Temperature | temp |
|
| Salinity | salt |
- Load Carbon Chemistry variables in the eReefs Data Explorer
- Learn more about Water Chemistry measures in eReefs
- Learn how to plot eReefs data on a map and plot an eReefs transect
Benthic Plants and Corals
All of the benthic plant and coral data variables are 2D! They can only occur on the seabed, so they do not have a depth dimension in the dataset. To work out the depth for these plants and corals occur at, you should look up the value of the botz coordinate variable at the same (i,j) grid indices.
CS_N), Halophilia Seagrass Nitrogen (SGH_N) and Deep Seagrass Nitrogen (SGD_N) from the GBR4_H2p0_B3p1_Cq3b_Dhnd dataset| Concept | Data Variable | Units |
|---|---|---|
| Coral Host Nitrogen | CH_N |
|
| Coral Symbiont Nitrogen | CS_N |
|
| Seagrass Zostera spp. Nitrogen | SG_N |
|
| Seagrass Halophila deciphens Nitrogen | SGD_N |
|
| Seagrass Halophila ovalis Nitrogen | SGH_N |
|
| Macroalgae Nitrogen | MA_N |
- Load Benthic Plants and Corals variables in the eReefs Data Explorer
- Learn more about macroalgae, seagrass and corals in eReefs
- Learn how to plot eReefs data variables on a map
Optics and Remote Sensing
Got Clouds? The eReefs Biogeochemistry and Sediments model results include variables for simulated satellite observations, including MODIS, VIIRS and the Sentinel-3 Ocean and Land Colour Instrument (OLCI).
Warning: Several of the optical variables have changed name in the newest version of the model (B4p2). Make sure you use the names that match the version of the dataset you are working with
| Concept | Data Variable | Units |
|---|---|---|
| Red: Simulated Sentinel-3 OLCI Remote Sensing Reflectance Band 08 (665nm). Used for Chl, sediment, vegetation |
B3p1: R_665B4p2: Sentinel_3B_B8 |
|
| Green: Simulated Sentinel-3 OLCI Remote Sensing Reflectance Band 06 (560nm). Used for Chlorophyll reference |
B3p1: R_560B4p2: Sentinel_3B_B6 |
|
| Blue: Simulated Sentinel-3 OLCI Remote Sensing Reflectance Band 04 (490nm). Used for high Chl, other pigments |
B3p1: R_490B4p2: Sentinel_3B_B4 |
GBR4_H2p0_B3p1_Cq3b_Dhnd dataset- Load Optics and Remote Sensing variables in the eReefs Data Explorer
- Learn more about simulated true colour in eReefs
- Learn how to plot simulated True Colour with eReefs data
Hydrodynamic Velocity Fields
We recommend using GBR1 Hydrodynamics datasets for working with velocity fields. For even better results, you may need to nest a higher-resolution model inside GBR1. (Talk to us about RECOM!)
(wspeed_u:wspeed_v), and Sea water velocity (u:v) at -0.5m and -31m from the GBR1_H2p0 dataset| Concept | Data Variable | Units |
|---|---|---|
| Sea Water Velocity: | ||
| - Eastward Current | u |
|
| - Northward Current | v |
|
| - Vertical Current | w |
|
| Sea Surface Wind: | ||
| - Eastward Wind | wspeed_u |
|
| - Northward Wind | wspeed_v |
- Load velocity fields in the eReefs Data Explorer
- Learn more about Sea water velocity currents in eReefs, and Current magnitude in eReefs
- Learn how to plot eReefs velocity fields on a map