The Area of Applicability (AOA) describes the area to which a predictive model can reliably be applied, based on the predictor space covered by the underlying training data. It was evaluated following the approach proposed by Meyer and Pebesma (2021).<br></br> The JRSRP seasonal surface reflectance composites between winter 2014 and winter 2024 were used as a proxy for the range of representative surface reflectance values likely to be encountered across the continent under varying environmental conditions from which fractional cover predictions are made. The AOA of the FCv3 model was computed for each seasonal surface reflectance composite, then summarised as a frequency map representing the proportion of seasons that a location was outside the AOA.<br></br> For each state, five files are provided: an annual product summarising the AOA across all seasons, and four showing seasonal AOA frequencies for summer, autumn, winter, and spring.<br></br>

<p>The dataset consists of composited seasonal surface reflectance images (4 seasons per year) created from the full time series of Landsat TM/ETM+/OLI imagery. The imagery has been composited over a season to produce imagery which is representative of that period, using techniques which will reduce contamination by cloud and other problems. This creates a regular time series of reflectance values which captures the variability at seasonal time scales. The benefits are a regular time series with minimal missing data or contamination from various sources of noise as well as data reduction. Each season has exactly one value (per band) for each pixel (or is null, i.e., missing), and the value for that season is assumed to be the representative of the whole season. The algorithm is based on the medoid (in reflectance space) over the time period (the medoid is a multi-dimensional analogue of the median), which is robust against extreme values.</p>

The seasonal dynamic reference cover method product compares the current ground cover level of each pixel to a reference pixel based on the historical timeseries and is available for Queensland from 1987 to present. It is created using a modified version of the dynamic reference cover method developed by <a href="https://doi.org/10.1016/j.rse.2012.02.021">Bastin et al (2012) </a>. This approach calculates a minimum ground cover image over all years to identify locations of most persistent ground cover in years with the lowest rainfall, then uses a moving window approach to calculate the difference between the window's central pixel and its surrounding reference pixels. The output is a difference image between the cover amount of a pixel's reference pixels and the actual cover at that pixel for the season being analysed. Negative values indicate pixels which have less cover than the reference pixels.<br> The main differences between this method and the original method are that this method uses seasonal fractional ground cover rather than the preceding ground cover index (GCI) and this method excludes cleared areas and certain landforms (undulating slopes), which are considered unsuitable for use as reference pixels.<br> This product is based upon the JRSRP Fractional Cover 3.0 algorithm.

An estimate of persistent green cover per season across Australia from 1989 to the present season, minus 2 years. This is intended to estimate the portion of vegetation that does not completely senesce within a year, which primarily consists of woody vegetation (trees and shrubs), although there are exceptions where non-woody cover remains green all year round. It is derived by fitting a multi-iteration minimum weighted smoothing spline through the green fraction of the seasonal fractional cover (dp1) time series. A single band image is produced: persistent green vegetation cover (in percent). The no data value is 255.

Version 1 of the Brigalow Belt Bioregion Spatial BioCondition dataset is superseded by the Version 2 dataset that can be found at: https://doi.org/10.25901/rnqz-cn10.<br><br> Version 1 was an initial demonstration version. The version 1 data has been removed from publication to negate temporal comparisons between v1 (2019) and v2 (2021), as this is a future goal for the product but still in development phase. This was a spatial dataset comprising predictions of vegetation condition for biodiversity for the brigalow belt bioregion. The dataset was created using a gradient boosting decision tree (GBDT) model based on eight vegetation specific remote sensing (RS) datasets and 17,000 training sites of known vegetation community and condition state. Condition score was modelled as a function of the difference in the RS space within homogeneous vegetation communities. The product was intended to represent predicted BioCondition for year 2019 rather than any single date.

Version 1 of the Southeast Queensland Bioregion Spatial BioCondition dataset is superseded by the Version 2 dataset that can be found at: https://doi.org/10.25901/r976-1v85.<br></br> Version 1 was an initial demonstration version. The version 1 data has been removed from publication to negate temporal comparisons between v1 (2019) and v2 (2021), as this is a future goal for the product but still in development phase. <br></br> This was a spatial dataset comprising predictions of vegetation condition for biodiversity for the Southeast Queensland Bioregion. The dataset was created using a gradient boosting decision tree (GBDT) model based on eight vegetation specific remote sensing (RS) datasets and 17,000 training sites of known vegetation community and condition state. Condition score was modelled as a function of the difference in the RS space within homogeneous vegetation communities. The product was intended to represent predicted BioCondition for year 2019 rather than any single date.

This dataset indicates the presence and persistence of water across Queensland between 1988 and 2022. Water is one of the world’s most important resources as it’s critical for human consumption, agriculture, the persistence of flora and fauna species and other ecosystem services. Information about the spatial distribution and prevalence of water is necessary for a range of business, modelling, monitoring, risk assessment, and conservation activities. The water count product is based on water index and water masks for Queensland (Danaher & Collett 2006) and represents the proportion of observations with water present across the Landsat time series as a fraction of total number of possible observations for the period 1 Jan 1988 to 31 Dec 2022. The product has two bands where band 1 is the number of times water was present across the time series, and band 2 is the count of unobscured (i.e. non-null) input pixels, or number of total observations for that pixel. Cloud, cloud-shadow, steep slopes and topographic shadow can obscure the ability to count water presence.

Foliage Projective Cover (FPC) is the percentage of ground area occupied by the vertical projection of foliage. The Remote Sensing Centre FPC mapping is based on regression models applied to dry season (May to October) Landsat-5 TM, Landsat-7 ETM+ and Landsat-8 OLI imagery for the period 1988-2014. An annual woody spectral index image is created for each year using a multiple regression model trained from field data collected mostly over the period 1996-1999. A robust regression of the time series of the annual woody spectral index is then performed. The estimated foliage projective cover is the prediction at the date of the selected dry season image for 2014. Where this deviates significantly from the woody spectral index for that date, further tests are undertaken before this estimate is accepted. In some cases, the final estimate is the woody spectral index value rather than the robust regression prediction. The product is further masked to remove areas classified as non-woody.

The data set is a statewide annual composite of fire scars (burnt area) derived from all available Landsat 5, 7 and 8 images acquired over the period January to December using time series change detection. Fire scars are automatically detected and mapped using dense time series of Landsat imagery acquired over the period 1987 - present. In addition, from 2013, products have undergone significant quality assessment and manual editing. The automated Landsat fire scar map products covering the period 1987-2012 were validated using a Landsat-derived data set of over 500,000 random points sampling the spatial and temporal variability. On average, over 80% of fire scars captured in Landsat imagery have been correctly mapped with less than 30% false fire rate. These error rates are significantly reduced in the edited 2013-2016 fire scar data sets, although this has not been quantified. <br> For the 2016 annual fire scar composite, the manual editing stage incorporated Landsat and Sentinel 2A imagery (resampled to match Landsat spatial resolution), allowing for increased cloud-free ground observations, and an associated reduction in the number of missed fires (not quantified). Sentinel 2A images were primarily used to map fire scars that were otherwise undetectable in the Landsat sequence due to cloud cover/Landsat revisit time. Additionally, Landsat-7 SLC-Off imagery (affected by striping) was excluded from the 2016 annual composite. It is expected that these modifications should result in improved mapping accuracy for the 2016 period.<br> A new fire scar detection algorithm has been developed, with a new edited product implemented in 2021.

Two fractional cover decile products, green cover and total cover, are currently produced from the historical timeseries of seasonal fractional cover images across Australia, available for each 3-month calendar season. These products compare, at the per-pixel level, the level of cover for the specific season of interest against the long term cover for that same season. For each pixel, all cover values for the relevant seasons within a baseline period (1990 - 2020) are classified into deciles. The cover value for the pixel in the season of interest is then classified according to the decile in which it falls.<br> This product is based upon the JRSRP Fractional Cover 3.0 algorithm.