Joint Remote Sensing Research Program & Department of the Environment, T. (2022). Seasonal Fractional Cover - Sentinel-2, JRSRP Algorithm Version 3.0, Eastern and Central Australia Coverage. Version 3.0. Terrestrial Ecosystem Research Network. Dataset. https://portal.tern.org.au/metadata/13810293-c6b5-442b-bfcd-817700738e0d
Seasonal Fractional Cover - Sentinel-2, JRSRP Algorithm Version 3.0, Eastern and Central Australia Coverage
Ver: 3.0
Status of Data: onGoing
Update Frequency: quarterly
Security Classification: unclassified
Record Last Modified: 2025-12-02
Viewed 17582 times
Accessed 711 times
Dataset Created: 2022-03-28
Dataset Published: 2022-05-03
Data can be accessed from the following links:
HTTPPoint-of-truth metadata URLWMSsentinel_fractional_v3HTTPLandscape Data Visualiser - Seasonal Fractional Cover - Sentinel-2, JRSRP Algorithm Version 3.0, Eastern and Central Australia CoverageHTTPDifferences between Fractional Cover version 2 and version 3HTTPro-crate-metadata.jsonHTTPVegmachine Timeseries ViewerHTTPCloud Optimised GeoTIFFs - Seasonal Fractional Cover - Sentinel-2, v3.0HTTPGitLab Code for Fractional Cover version 3
How to cite this collection:
The seasonal fractional cover product shows representative values for the proportion of bare, green and non-green cover, created from a time series of Sentinel-2 imagery. It is a spatially explicit raster product, which predicts vegetation cover at medium resolution (10 m per-pixel) for each 3-month calendar season across Eastern and Central Australia from 2016 to present. The green and non-green fractions may include a mix of woody and non-woody vegetation. This model was originally developed for Landsat imagery, but has been adapted for Sentinel-2 imagery to produce a 10 m resolution equivalent product. A 3 band (byte) image is produced:
- band 1 - bare ground fraction (in percent),
- band 2 - green vegetation fraction (in percent),
- band 3 - non-green vegetation fraction (in percent).
Credit
We at TERN acknowledge the Traditional Owners and Custodians throughout Australia, New Zealand and all nations. We honour their profound connections to land, water, biodiversity and culture and pay our respects to their Elders past, present and emerging. This dataset was produced by the Joint Remote Sensing Research Program using data sourced from the European Space Agency (ESA) Copernicus Sentinel Progam.
Purpose
This product captures variability in fractional cover at seasonal (i.e. three-monthly) time scales, forming a consistent time series from late 2015 - present. It is useful for investigating recent inter-annual changes in vegetation cover and analysing regional comparisons. For applications that focus on non-woody vegetation, the Landsat-derived ground cover product may be more suitable. For applications investigating rapid change during a season, the monthly composite or single-date (available on request) fractional cover products may be more appropriate. This product is based upon the JRSRP Fractional Cover 3.0 algorithm.
Lineage
Summary of processing:
Sentinel 2 surface reflectance data > multiple single-date fractional cover datasets > seasonal composite of fractional cover
Further details are provided in the Methods section.
Sentinel 2 surface reflectance data > multiple single-date fractional cover datasets > seasonal composite of fractional cover
Further details are provided in the Methods section.
Method DocumentationZhu, Z., & Woodcock, C. E. (2012). Object-based cloud and cloud shadow detection in Landsat imagery. Remote Sensing of Environment, 118, 83–94. https://doi.org/10.1016/J.RSE.2011.10.028Flood, N. (2013). Seasonal Composite Landsat TM/ETM+ Images Using the Medoid (a Multi-Dimensional Median). Remote Sensing, 5(12), 6481–6500. https://doi.org/10.3390/rs5126481Flood, N. (2017). Comparing Sentinel-2A and Landsat 7 and 8 Using Surface Reflectance over Australia. Remote Sensing, 9(7). https://doi.org/10.3390/rs9070659Sentinel 2 Level 1C Algorithms and ProductsMuir, J., Schmidt, M., Tindall, D., Trevithick, R., Scarth, P., & Stewart, J. B. (2011). Field measurement of fractional ground cover: A technical handbook supporting ground cover monitoring for Australia.
Procedure Steps
1.
Image Pre-processing:
Sentinel-2 data was downloaded from the ESA as Level 1C (version 02.04 system). Masks for cloud, cloud shadow, topographic shadow and water were applied as described in Flood (2017).
Sentinel-2 data was downloaded from the ESA as Level 1C (version 02.04 system). Masks for cloud, cloud shadow, topographic shadow and water were applied as described in Flood (2017).
2.
Fractional Cover Model:
A multilayer perceptron (MLP) model is used to estimate percentage cover in three fractions – bare ground, photosynthetic vegetation (PV) and non-photosynthetic vegetation (NPV) from surface reflectance, for every image captured within the season. The MLP model was trained with Tensorflow using Landsat TM, ETM+ and OLI surface reflectance and a collection of approximately 4000 field observations of overstorey and ground cover. The field observations covered a wide variety of vegetation, soil and climate types across Australia, collected between 1997 and 2018 following the procedure outlined in Muir et al (2011). As the model is trained on Landsat imagery, the Sentinel-2 reflectance values are slightly adjusted to more closely resemble Landsat imagery, then the fractional cover model is applied. The model was assessed to predict the vegetation cover fractions with MAE/wMAPE/RMSE of:
bare - 6.9%/34.9%/14.5%
photosynthetic vegetation (PV) - 4.6%/37.9%/10.6%
non-photosynthetic vegetation (NPV) - 9.8%/25.2%/16.9%.
A multilayer perceptron (MLP) model is used to estimate percentage cover in three fractions – bare ground, photosynthetic vegetation (PV) and non-photosynthetic vegetation (NPV) from surface reflectance, for every image captured within the season. The MLP model was trained with Tensorflow using Landsat TM, ETM+ and OLI surface reflectance and a collection of approximately 4000 field observations of overstorey and ground cover. The field observations covered a wide variety of vegetation, soil and climate types across Australia, collected between 1997 and 2018 following the procedure outlined in Muir et al (2011). As the model is trained on Landsat imagery, the Sentinel-2 reflectance values are slightly adjusted to more closely resemble Landsat imagery, then the fractional cover model is applied. The model was assessed to predict the vegetation cover fractions with MAE/wMAPE/RMSE of:
bare - 6.9%/34.9%/14.5%
photosynthetic vegetation (PV) - 4.6%/37.9%/10.6%
non-photosynthetic vegetation (NPV) - 9.8%/25.2%/16.9%.
3.
Data Compositing:
The method of compositing selected representative pixels through the determination of the medoid (multi-dimensional equivalent of the median) of at least three observations of fractional cover imagery. The medoid is the point which minimises the total distance between the selected point and all other points. Thus the selected point is “in the middle” of the set of points. It is robust against extreme values, inherently avoiding the selection of outliers, such as occurs when cloud or cloud shadow goes undetected. Unfortunately, due to the high level of cloud cover in some areas, often three cloud free pixels are not available, resulting in data gaps in the seasonal fractional cover image. For further details on this method see Flood (2013).
The method of compositing selected representative pixels through the determination of the medoid (multi-dimensional equivalent of the median) of at least three observations of fractional cover imagery. The medoid is the point which minimises the total distance between the selected point and all other points. Thus the selected point is “in the middle” of the set of points. It is robust against extreme values, inherently avoiding the selection of outliers, such as occurs when cloud or cloud shadow goes undetected. Unfortunately, due to the high level of cloud cover in some areas, often three cloud free pixels are not available, resulting in data gaps in the seasonal fractional cover image. For further details on this method see Flood (2013).
Australia excluding Western Australia and South Australia
Temporal Coverage
From 2015-12-01 to on going
Spatial Resolution
Vertical Extent
Data not provided.
Data Quality Assessment Scope
Sentinel 2 Performance and Data Quality Reports1) All the data described here has been generated from the analysis of Sentinel-2 data, which has a spatial resolution of approximately 10 m in the Blue, Green, Red and Near Infra-red (NIR) bands, and 20 m in the two Short Wave Infra-red (SWIR) band. The 20 m bands have been resampled to 10 m using cubic convolution, to provide a consistent 10 m data set. The imagery is rectified during processing by the European Space Agency (ESA), and not modified spatially beyond that.
2) The fractional cover model was compared to samples drawn from approximately 4000 field reference sites.
2) The fractional cover model was compared to samples drawn from approximately 4000 field reference sites.
Data Quality Assessment Outcome
1) The Sentinel-2 Data Quality Report from ESA indicates that positional accuracy is on the order of 12 m.
2) The fractional cover model predicts the vegetation cover fractions with MAE/wMAPE/RMSE of:
bare - 6.9%/34.9%/14.5%
PV - 4.6%/37.9%/10.6%
NPV - 9.8%/25.2%/16.9%.
2) The fractional cover model predicts the vegetation cover fractions with MAE/wMAPE/RMSE of:
bare - 6.9%/34.9%/14.5%
PV - 4.6%/37.9%/10.6%
NPV - 9.8%/25.2%/16.9%.
Horizontal Resolution
Instruments
Parameters
Platforms
Temporal Resolution
Supplemental Information
Data are available as cloud optimised GeoTIFF (COG) files. COG files are easier and more efficient for users to access data corresponding to particular areas of interest without the need to download the data first.
Resource Specific Usage
Data not provided.
Environment Description
Data not provided.
Terrestrial Ecosystem Research Network
80 Meiers Road, Indooroopilly, Queensland, 4068, Australia.
Contact Us
TERN services are provided on an “as-is” and “as available” basis. Users use any TERN services at their discretion and risk. They will be solely responsible for any damage or loss whatsoever that results from such use including use of any data obtained through TERN and any analysis performed using the TERN infrastructure.
Web links to and from external, third party websites should not be construed as implying any relationships with and/or endorsement of the external site or its content by TERN.
Please advise any work or publications that use this data via the online form at https://www.tern.org.au/research-publications/#reporting
Web links to and from external, third party websites should not be construed as implying any relationships with and/or endorsement of the external site or its content by TERN.
Please advise any work or publications that use this data via the online form at https://www.tern.org.au/research-publications/#reporting
It is not recommended that these data sets be used at scales more detailed than 1:100,000.