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 US Geological Survey.
Purpose
This product captures variability in persistent green cover at seasonal (i.e. three-monthly) time scales, forming a consistent time series from 1989 to the present season (minus 2 years). It is useful for investigating inter-annual and longer term changes in persistent vegetation cover. The statistical process used to create this product means there is a 2-year lag in producing it.
For applications that focus on non-woody vegetation, the ground cover product, derived from fractional cover, may be more suitable.
For applications investigating rapid change during a season, 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.
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.
File naming convention:
Filenames for the seasonal persistent green product conform to the AusCover standard naming convention. The standard form of this convention is:
{satellite category code}{instrument code}{product code}_{where}_{when}_{processing stage code}_{additional dataset specific tags}
Details of the codes used for this dataset can be found in the attached txt file: File Naming Convention.
Raster data values represent percentages of persistent green cover, with a no data value of 255.
Lineage
Summary of processing:
Landsat surface reflectance data > multiple single-date fractional cover datasets > seasonal composite of fractional cover > seasonal persistent green product.
Data Creation
Image Pre-Processing:
All input Landsat TM/ETM+/OLI imagery was downloaded from the USGS EarthExplorer website as level L1T imagery. Images which the EarthExplorer site rated as having greater than 80% cloud cover were not downloaded. The imagery has been corrected for atmospheric effects, and bi-directional reflectance and topographic effects, using the methods detailed by Flood et al (2013). The result is surface reflectance standardised to a fixed viewing and illumination geometry. Cloud, cloud shadow and snow have been masked out using the Fmask automatic cloud mask algorithm. Topographic shadowing has been masked using the Shuttle Radar Topographic Mission DEM at 30 m resolution. Water has been masked out using the methods outlined in Danaher & Collett (2006).
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). The model was assessed to predict the vegetation cover fractions with MAE/wMAPE/RMSE of:
bare - 6.9%/34.9%/14.5%
PV - 4.6%/37.9%/10.6%
Seasonal Compositing:
The method of compositing used selection of representative pixels through the determination of the medoid (multi-dimensional equivalent of the median) of three months (a season) 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. The value selected is a specific data point and not an averaged or blended value. It is robust against extreme values, inherently avoiding the selection of outliers, such as occurs when cloud or cloud shadow goes undetected. At least three pixels from the time-series of imagery for the season must be available. 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).
Persistent Green Fractional Cover:
Smoothing splines are fitted in multiple iterations per pixel through the full time series of seasonal fractional cover (green fraction only). At each iteration, zero weight is given to observations that lie above the spline, and observation below the line are weighted proportion to the size of the residual. Observations greater than 3 standard deviations from the residual mean are given zero weight, and those between 2 and 3 standard deviations are given less weight, this avoids contamination by outliers. Persistent green fractional cover for each season is estimated from the final spline iteration at each seasonal time step. Values reported are as for fractional cover, ie. percentages of cover.
Areas with frequent seasonal fractional cover data gaps due to cloud may produce unreliable estimates of persistent green cover.
A single band (byte) image is produced: persistent green vegetation cover (in percent).
Authors
Point of Contact
Publisher
Rights Holder
Data Quality