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 ground cover at seasonal (ie three-monthly) time scales, forming a consistent time series from 1989 - present. It is useful for investigating inter-annual changes in ground cover and analysing regional comparisons. The green and non-green fractions may include a mix of woody and non-woody vegetation. For applications that focus on all vegetation, the fractional cover product 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. Note: A new fractional cover algorithm will be implemented during 2021, based on additional field validation and a new machine learning approach. This will lead to a new version of the ground cover products.
Supplemental Information
band 1 - bare ground fraction (in percent) + 100 band 2 - green ground cover fraction (in percent) +100 band 3 - non-green ground cover fraction (in percent) + 100 band 4 - Error Layer representing the RMSE between the predicted pixel value and the actual pixel value on a nominal scale of 100 (no error) to 200 (very large error). The standard form of the file naming convention is: _ _ _ _ For this dataset are (comma-separated): Data Naming Element, Possible Code(s), Descriptor Standard Elements satellite category, lz, Landsat - all possible instrument, tm, thematic product, re, reflective where, qld,nsw,vic,tas,sa,nt,wa, state when , myyyymmyyyymm , season start date (1st day of month) and season end date (last day of month) processing stage, dix, seasonal fractional ground cover data projection, a2, Australian Albers Equal Area
Lineage
Data Creation
Algorithm Summary 1:
When viewed from above, as by the satellite, mid- and over-storey vegetation obscures the ground layer fractions (bare ground, green ground cover and dry ground cover). The seasonal fractional cover product, from which this ground cover product is derived, does not distinguish between these vegetation layers. As such, a high estimate in the green fraction may be due to green vegetation in any of the vegetation layers.
To produce the ground cover product, the fractional cover product is adjusted using an estimate of the proportion of the pixel obscured by mid- and over-story foliage. This estimate of cover is an estimate of the combined persistent dry and persistent green layers. That is, all vegetation in the mid- and upper- stories. The mid and upper storage foliage is effectively removed from the estimates of cover and the ground cover estimate is therefore based only on the proportion of ground that was visible by the satellite.
Algorithm Summary 2:
We assume that fractional cover fractions in each vegetation layer are independent, that is that the presence of mid and over-storey vegetation does not influence the distribution of the ground cover fractions.
The algorithm to create the ground cover layer relies on two satellite products (seasonal fractional cover and seasonal persistent green), as well as an estimate of persistent dry derived from field data.
A simplified description of the algorithm follows:
1. Base imagery required is seasonal fractional cover and seasonal persistent green products.
2. Estimate of persistent dry is derived from field data relationship
3. The ‘gap-fraction’ is calculated (total pixel area - pixel area composed of persistent dry - pixel area composed of persistent green)
4. Adjust all three fractions for each pixel using persistent dry and persistent green estimates
Base Imagery:
All seasonal ground cover images are derived directly from the equivalent date seasonal fractional cover image. See the seasonal fractional cover product metadata for details. Persistent green images can only be created retrospectively, so when the most current ground cover images are created using the most recent persistent green, whatever this may be. These images are tagged ‘_vinterim’ to indicate they will eventually be replaced, when the exact date persistent green becomes available. It is expected that the interim images will be almost identical in nature to the final products when they become available.
Fractional Cover:
Land cover fractions representing the proportions of green, non-green and bare cover retrieved by inverting multiple linear regression estimates and using synthetic endmembers in a constrained non-negative least squares unmixing model.
The opening of the Landsat archive has provided an opportunity to composite imagery into representative seasonal images. The benefits of compositing in this manner are the creation of a regular time-series capturing seasonal variability, and the minimisation of missing data and contamination present in single date imagery (Flood, 2013). Further details of the product can be found on the seasonal fractional cover product page.
Persistent Green:
The persistent green model estimates persistent green by investigating the long term green fraction of the fractional cover product and determining the minimum green that is present regardless of seasonality. The underlying premise of the persistent green product is the separation of the fractional cover product green fraction into variable and trend components. The trend component is the green fraction which is always present regardless of season and is therefore assumed to be associated with perennial vegetation and therefore ‘persistent’.
Persistent Dry:
The adjustments made rely both on satellite estimates of persistent green and persistent dry, as the calculation of total gap depends on this. While we currently have satellite estimates of persistent green vegetation, we do not have satellite estimates of persistent dry. We can, however, derive a relationship between persistent dry and persistent green from the field data. This relationship can then be used to estimate persistent dry from the satellite persistent green product.
Canopy Gap Fraction:
The visible mid FPC is the proportion of total mid-level green measurements multiplied by the proportion of the site you can see because it is not excluded by canopy leaves, branches or dead vegetation (canopy gap). The canopy gap is calculated from the field data as the total number of points observed in the over-storey divided by the total number of observations taken at the site. The mid FPC is the total mid storey green observations divided by the total number of observations at the site.
Data Storage:
A 3 band (byte) image is produced: band 1 – bare ground fraction (in percent) + 100. band 2 - green vegetation fraction (in percent) +100. band 3 – non-green vegetation fraction (in percent) + 100
Authors
Point of Contact
Publisher
Data Quality