This dataset and associated data services provide access to modelled 3D environmental conditions of the waters of the Great Barrier Reef and Coral Sea regions. They correspond to generated summaries (daily, monthly, annual, overall) of the eReefs CSIRO hydrodynamic model (v2.0 and v4.0) at both 1km and 4km resolution, along with associated visualisation products.
The base eReefs models provide hydrodynamic data with an hourly time interval. This makes the data very large, requiring 630 MB / day for the 4 km grid data and 9.1GB / day for the 1 km grid. Where analysis only requires a small subset of the modelled area the OpenDAP subsetting services on NCI can be used to extract the raw high temporal resolution data. However, where the analysis needs to cover large areas over long periods of time the huge data size can be a significant problem. For many types of analyses daily, or monthly averages provide sufficient temporal resolution. This temporal averaging drastically reduces the total data size, allowing analyses over long periods. This dataset provides a range of precomputed temporal aggregation averages (daily, monthly, annual) of the eReefs hydrodynamic models. We also provide a visualisation of the average over the entire time series (Overall).
Another challenge that this dataset aims to address is the difficultly of working with the curvilinear grid used by the base eReefs models. Curvilinear grids warp the model grid to more closely align with the curve of the Queensland coastline, allowing the model to be more computationally efficient. Unfortunately, not many tools know how to correctly read the curvilinear grids, making it difficult to work with the base eReefs model files. In this dataset the original curvilinear data has been regridded onto a regular rectangular grid, allowing the files to be easily loaded and processed with a wide range of GIS tools and programming libraries.
These products are generated by the AIMS eReefs Platform (https://ereefs.aims.gov.au/ereefs-aims). The data products are available as an OPeNDAP service and as NetCDF files (https://thredds.ereefs.aims.gov.au/thredds/). These summaries are derived from the original hourly model outputs available via the National Computing Infrastructure (NCI) (https://thredds.nci.org.au/thredds/catalog/catalogs/fx3/catalog.html).
v4.0 hydrodynamic model:
The v4.0 hydrodynamic model is a hindcast model that is periodically updated with an extended time range every couple of years. As a hindcast this model is driven from the best available input drivers that are consistent over the full modelled time period. This is a significant change from v2.0 of the models that were updated in near real time.
v2.0 model (deprecated):
Version 2.0 of the hydrodynamic model was updated in near-real time, with the model outputs typically being available with 1 week of the current time. While this allowed the model to useful for understanding current events the download side was the model could only be driven from a small pool of near-time sources that relied on automated instrumentation with little quality control. Another limitation of this model was that as various input data streams came and went over time the source input data streams varied over time. This meant the model is not entirely consistent over its full time period.
The near-real time model was stopped in Jan 2024 when several river gauge stations were damaged due to cyclone Japser.
Notice: In 2026 we may remove the deprecated v2.0 eReefs models from our Thredds services to save on storage costs. They will be migrated to archival storage where they will only be available on request. If you are using this version it is recommended that you retain a copy of the portion used in your research for reproducibility purposes.
Method:
A description of the processing, especially aggregation and regridding, is available in the "Technical Guide to Derived Products from CSIRO eReefs Models" document (https://nextcloud.eatlas.org.au/apps/sharealias/a/aims-ereefs-platform-technical-guide-to-derived-products-from-csiro-ereefs-models-pdf ). For the GBR 4 model the outputs were re-gridded to 0.03 degrees (~3 km) and the GBR 1 model is regridded to 0.005 degrees (~500 m). The regular grids were chosen to be finer than the original grids to minimise defects introduced by the regridding process.
Data Dictionary:
The following variables are available:
- eta: Surface elevation (sea surface height above sea level) (metres) - This can be used to estimate tides.
- salt: Salinity (PSU)
- temp: Temperature (degrees C)
- wspeed_u: Eastward wind (ms-1) - Average over time
- wspeed_v: Northward wind (ms-1) - Average over time
- u: Eastward sea water velocity (component of the ocean current vector in the eastward direction) (ms-1)
- v: Northward sea water velocity (component of the ocean current vector in the northern direction) (ms-1)
- mean_wspeed: Mean Wind speed magnitude - Average of the hourly wspeed magnitude over time (ms-1)
- mean_cur: Mean sea water velocity (current) magnitude - Average of the hourly current magnitude over time (ms-1)
FAQ:
With the aggregate results why is the mean_cur not equal to the sqrt(u^2+v^2)?
The mean_cur is not equal to the sqrt(u^2 + v^2) in the aggregate results, because the u and v represent the net current flow, with most of the tidal signal averaged out, while the mean_cur represents the mean of the current magnitude which includes tidal flows.
With the instantaneous (raw hourly) model data current_magitude=sqrt(u^2 + v^2). After temporal aggregation (this dataset) u and v are an average over time of the vector components. The positive and negative oscillations in the original hourly values, typically due to the tides, tend to be average to zero over time. As a result the temporal aggregate u and v components represent the vector of the net current flow. In high tidal areas the net flow can be quite low, with water sloshing back and forth to the same location, even though the average of the magnitude of the current is high (mean_cur).
The mean_cur is therefore useful for understanding bottom steer stress and mixing due to currents, while the temporal aggregate u and v components are useful for understanding large ocean currents.
How does the regridding work? What algorithm is used?
The raw eReefs model (available via NCI) use a curvilinear grid to minimise the number of simulation cells. This grid format is incompatible with many GIS applications. The products associated with this dataset, available from the AIMS eReefs THREDDS server, are regridded on to a regular rectangular grid. This regridding is performed using an Inverse Distance Weight from the nearest 4 grid cells. A consequence of this approach is that there are an additional set of interpolated pixels along the coastline. Additional detail on the regridding is available in the "Technical Guide to Derived Products from CSIRO eReefs Models".
Depths:
This data set contains a subset of the depths available in the source data sets, which differ slightly between the 1km and 4km models.
Depths at 1km resolution: -0.5, -2.35, -5.35, -9.0, -13.0, -18.0, -24.0, -31.0, -39.5, -49.0, -60.0, -73.0, -88.0, -103.0, -120.0, -140.0,
Depths are 4km resolution:-0.5, -1.5, -3.0, -5.55, -8.8, -12.75, -17.75, -23.75, -31.0, -39.5, -49.0, -60.0, -73.0, -88.0, -103.0, -120.0, -145.0.
Limitations:
The wind data is originally from the BOM Access-R weather models. These models capture synoptic winds and some of the features of cyclones, however they do not represent the high speed winds near the eye of cyclones well. For this reason the maximum wind speed aggregations do not capture the peak winds of cyclones.
The GBR1 yearly summaries of salinity and temperature have a known problem where the data is WRONG with a sharp boundary appearing at edge of the GBR. Do not use this data. We will investigate and resolve the issue with this data.
This dataset is based on a spatial and temporal model and as such is an estimate of the environmental conditions. It is not based on in-water measurements, and thus will have a spatially varying level of error in the modelled values. It is important to consider if the model results are fit for the intended purpose.
Change Log:
2020-08-18 (Version 1): Initial version of the system and dataset. Dataset and service covers GBR4 BGC v3.1 from 2010 - 2019.
2025-05-16 (Version 1): Improved the metadata title and abstract to better represent the dataset. Also minted a DOI and added how the dataset should be cited.
2025-10-16 (Version 2): Added the description for the v4.0 model data. Expanded the description to include the visualisation products. Added description for 'Overall' product. Updated, now deprecated, links to the NCI Thredds.
