AusEFlux: Empirical upscaling of OzFlux eddy covariance flux tower data over Australia

AusEFlux (Australian Empirical Fluxes) is a high resolution (500 metre) gridded estimate of Gross Primary Productivity (GPP), Ecosystem Respiration (ER), Net Ecosystem Exchange (NEE), and Evapotranspiration over the Australian continent for the period January 2003 to Present.  These datasets provide a benchmark for assessment against Land Surface Model simulations, and a means for monitoring of Australia’s terrestrial carbon cycle at an unprecedented high-resolution.

Version 2.1 of AusEFlux has just been released (as of May 2025) and was created to operationalise the research datasets published in this EGU Biogeosciences publication. In order to operationalise these datasets, changes to the input datasets were required to align the data sources with datasets that are regularly and reliably updated, along with general improvements. The datasets provided on Zenodo have been reprojected to 5 km resolution to facilitate easier uploading and sharing, but full resolution datasets (both v1.1 and v2.1) can be accessed freely through NCI's THREDDS portal.

Two Jupyter Notebooks have been created (one for GPP and one for NEE) that demonstrate the differences between the research datasets (v1.1) and the operational datasets (v2.1), including showing the differences in specifications and inputs.  You can view/download these notebooks using the links below:

GPP comparison between versions

NEE comparisons between versions

Each dataset contains three variables:

• "<flux>_median": represents the 'best-estimate' of a given flux, the units are gC/m^₂/mon^-1
• "<flux>_25th_percentile": represents the lower uncertainty bound, the units are gC/m^₂/mon^-1
• "<flux>_75th_percentile": represents the upper uncertainty bound, the units are gC/m^₂/mon^-1

Version Guide:

v1.0: DO NOT USE THIS VERSION. There was a mistake in the modelling of ecosystem respiration, so this version of the dataset should not be used.  As of version 1.1, the error has been rectified.

v1.1: This version of the datasets are those used to inform the EGU Publication linked above. Its time range is 2003-July 2022, and its spatial resolution is 5 km on Zenodo, but the 1 km resolution datasets can be accessed through NCI's THREDDS portal.

v2.0: IMPORTANT NOTE: a bug in the modelling of vegetation height resulted in data artefacts in the NEE and ER fluxes over very tall mesic forests in this version. This resulted in unnaturally high ER and lower than expected NEE (less negative than would be expected). This issue has been rectified in version 2.1.  Version 2 datasets represent the operational version of the datasets, it includes several improvements over version 1.1. Its time-range is 2003-2024 (and will be updated annually), and its spatial resolution is 500m.  A 5 km reprojected version of the dataset is included here on Zenodo, but the 500 metre datasets can be accessed through NCI's THREDDS portal.

v2.1: This version is a patch to version 2.0 to remove a bug in the modelling of vegetation height. It is recommended to use this version over v2.0. 500 metre resolution datasets can be accessed through NCI's THREDDS portal.