Fire severity classification of bushfires (wildfires) impacting ~1.5 million hectares of predominantly forested public land in eastern and north-eastern Victoria (and ~300,000 ha of southern NSW), between November 2019 and March 2020. Fire severity mapping was derived using machine learning classification (Random forests) of eight Spectral Indices (SI) from pre and post fire Sentinel 2 satellite imagery. The fire severity classification model was trained using high resolution ( 20% canopy foliage consumed); ii) High canopy scorch (5) - HCS (>80% of canopy foliage is scorched); Medium canopy scorch (4) - MCS (Canopy is a mosaic of both unburnt and scorched foliage, 20 - 80%); iii) Low canopy scorch (3) - LCS (Canopy foliage is largely unaffected (90%) unburnt). Additional classes: v) No Data (0) (e.g. due to obscuration by cloud, cloud-shadow and/or smoke and haze) and vi) Non-woody vegetation (unclassified) (1).
An independent cross-validation of the classification model was used to estimate global and per-class model accuracy. Overall accuracy is estimated to be 85% (0.81 Kappa), with producer per-class accuracy ranging from 97% (CB), 91% (HCS), 88% (UB), 75% (LCS) and 61% (MCS). A ground-based validation of the classification has not been undertaken.
Data is provided on the basis that users undertake responsibility for assessing the relevance and accuracy of its content. The Department of Environment, Land, Water and Planning, on behalf of the Victorian Government, makes no representations, either expressed or implied, as to the suitability of this data for any particular purpose.
We do not accept any liability to any person:
- for the information, data or advice (or the use of such information, data or advice) which is provided or incorporated into it by reference
- for any interference with or damage to a user's computer, software or data occurring in connection with or relating to this data or its use
Related research can be found in L. Collins, P. Griffioen, G. Newell, A. Mellor (2018), The utility of Random Forests for wildfire severity mapping, Remote Sensing of Environment, 216, 374-384