1.0.0: DEA Burnt Area Vectors Sentinel-2 Near Real-Time

1.0.0: DEA Burnt Area Vectors Sentinel-2 Near Real-Time

DEA Burnt Area Vectors Sentinel-2 Near Real-Time

Version:

1.0.0 (See latest version)

Product types:

Derivative, Vector

Update frequency:

Deprecated

../../../_images/dea-hero.jpg

Old version

This is an old version of the product. See the latest version.

Provisional product

This is a provisional product, meaning it has not yet passed quality control and/or been finalised for release.

Access the data

For help accessing the data, see the Access tab.

Digital Earth Australia - Public Data (DEV)

Get the data online

Key details

Collection

Geoscience Australia Sentinel-2 Collection 3

Background

Bushfires pose a serious and increasing threat to Australia. Detection and mapping of burns have many applications to support areas impacted by fire. However, the identification of bushfire burn using Earth Observation is often manual, can come with a significant time delay, and at a relatively small scale. This product offers provisional and preliminary change detection using same day satellite data to automatically and rapidly identify burn characteristics.

Knowledge about the potential location and extent of fire helps to understand community and ecosystem impacts, enables directed relief and recovery support, and informs planning of mitigation burning for future fire seasons.

What this product offers

DEA Burnt Area Vectors contribute to the understanding of the distribution and frequency of fire, in the Australian continent, by identifying change in vegetation cover and soil characteristics that may be indicative of fire activity in the landscape. This product contains vector outlines identifying areas of change between a baseline image and the most recent observation on the Sentinel 2 satellite constellation.

Change is detected by combining the three indexes from the DEA Burnt Area Characteristic Layers, which describe characteristics of the earth surface that change as the result of a burn. This dataset identifies areas where the input Characteristic Layers detect substantial change, delivered as a vector outline product. This dataset also includes information on areas of the landscape that were not observed within the observation area. These areas might be smoke or cloud obscuring the satellite pass.

Applications

  • As a preliminary use-case to show the potential of near real-time change detection

  • As a screening tool to identify the potential location of new burnt areas

  • As a screening tool to identify the potential size of burnt areas

  • To visually identify potential changes of known burnt areas between two time periods

Technical information

This Near Real-Time (NRT) change detection product is based on the DEA Burnt Area Characteristic Layers. The product identifies areas of substantial change by combining the three change layers from the DEA Burnt Area Characteristic Layers; Delta Normalised Burn Ratio (\(\Delta\)NBR), Delta Bare Soil Index (\(\Delta\)BSI) and the Delta Normalised Difference Vegetation Index (\(\Delta\)NDVI).

Substantial change is defined by applying a threshold to each Characteristic Layer of + 0.1. The thresholded layers are then combined in the following way; areas where thresholded delta NBR (\(\Delta\)NBR) agrees with either of the other thresholded delta layers are considered to be possible burnt area. This possible burnt area then undergoes a process of binary dilation and erosion in order to remove noise from the dataset by removing outlying single 10 x 10 m pixels before areas of possible burn are converted into vector shapes.

A Data Quality Mask is generated from the Sentinel-2 Near Real-Time product fmask layer. Cloud, cloud shadow and water that are detected by fmask are used to generate a binary layer which undergoes the same dilation and erosion process as the possible burnt area. This layer is then converted into vector outlines which are included in the vector product as a Data Quality Mask indicating the areas which should be considered as “Not Analysed” by the change detection algorithms.

Lineage

The DEA Burnt Area Vectors are generated from the DEA Burnt Area Characteristic Layers and the Sentinel-2 (A and B combined) Near Real-Time provisional satellite data.

The DEA Burnt Area Characteristic Layers are combined to find areas of substantial change. These areas are converted into vectors before being masked with the fmask layer from the corresponding Sentinel-2 Near Real-Time scene.

References

Cocke A. E., Fulé P. Z., Crouse J. E. (2005) Comparison of burn severity assessments using Differenced Normalized Burn Ratio and ground data. International Journal of Wildland Fire 14, 189-198. https://doi.org/10.1071/WF04010

Dindaroglu, T., Babur, E., Yakupoglu, T., Rodrigo-Comino, J. and Cerdà, A., (2021) Evaluation of geomorphometric characteristics and soil properties after a wildfire using Sentinel-2 MSI imagery for future fire-safe forest. Fire Safety Journal (122). https://doi.org/10.1016/j.firesaf.2021.103318.

Huete, A.R., Jackson, R.D., (1987) Suitability of spectral indices for evaluating vegetation characteristics on arid rangelands, Remote Sensing of Environment, Volume 23( 2), (213), https://doi.org/10.1016/0034-4257(87)90038-1

Jin, S., Yang, L., Danielson, P., Homer, C., Fry, J. and Xian, G. (2013). A comprehensive change detection method for updating the National Land Cover Database to circa 2011. Remote Sensing of Environment 132. (159-175). https://doi.org/10.1016/j.rse.2013.01.012

Keeley, J. E. (2009). Fire intensity, fire severity and burn severity: A brief review and suggested usage. International Journal of Wildland Fire, 18(1), 116–126. https://doi.org/10.1071/WF07049

Krause, C. E., Newey, V., Alger, M, J., Lymburner, L., (2021). Mapping and Monitoring the Multi-Decadal Dynamics of Australia’s Open Waterbodies Using Landsat, Remote Sensing. 13(8), 1437. https://doi.org/10.3390/rs13081437

Rahman, S., Chang, H.C., Hehir, W., Magilli, C. and Tomkins, K., (2018) Inter-Comparison of Fire Severity Indices from Moderate (Modis) and Moderate-To-High Spatial Resolution (Landsat 8 & Sentinel-2A) Satellite Sensors. In IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium (2873-2876). https://doi.org/10.1109/IGARSS.2018.8518449

Rikimaru, A., Roy, P.S. and Miyatake, S., (2002). Tropical forest cover density mapping. Tropical ecology, 43(1), (39-47). https://www.tropecol.com/pdf/open/PDF_43_1/43104.pdf

Roberts, D., Wilford, J. & Ghattas, O. (2019). Exposed soil and mineral map of the Australian continent revealing the land at its barest. Nature Communications 10, Article 5297. https://doi.org/10.1038/s41467-019-13276-1

Rouse, J.W., Haas, R.H., Schell, J.A. and Deering, D.W., (1974). Monitoring vegetation systems in the Great Plains with ERTS. NASA special publication, 351(1974), (309).

United Nations, Normalized Burn Ratio (NBR), Office for Outer Space Affairs UN-SPIDER Knowledge Portal. https://un-spider.org/advisory-support/recommended-practices/recommended-practice-burn-severity/in-detail/normalized-burn-ratio

van Wagtendonk, J.W., Root, R.R., Key, C. H., (2004) Comparison of AVIRIS and Landsat ETM+ detection capabilities for burn severity, Remote Sensing of Environment, 92 (3), (397-408) https://doi.org/10.1016/j.rse.2003.12.015

Accuracy

There are a number of limitations when using this dataset to locate or map burnt area. There are limitations that apply to all Earth Observations with satellites and those specific to this product.

Limitations of Earth Observations

Observation of the landscape by this service depends on having clear skies. Both cloud and smoke will obstruct the observation of the Earth’s surface. Furthermore, the satellites do not observe all places all of the time. The two Sentinel 2 satellites (A and B), which are the basis for this service, each view a given 320 kilometre wide strip of Australia only once every 10 days. This means there will be up to a five day gap between observations of any point on the Australian continent. The observations show only what was visible on the day of the satellite pass and at the time it is overhead.

Fires are fast moving and the situation on the ground can change rapidly. No decisions on life or property should be made based on this data. For local updates and alerts, please refer to your state emergency or fire service.

Limitations specific to this dataset

A number of changes in the landscape with similar characteristics to burn appear in this dataset. These include but are not limited to:

  • Forestry harvesting

  • Land clearing

  • Vegetation dieback

  • Extreme dry conditions during a drought

  • Increase in water body area

A number of factors may contribute to a burnt area from an actual fire not being visible in this product. Potential situations in which a burnt area may appear to have very little or no change from the baseline state are as follows:

  • A burn in shrubland or grassland that experience a seasonal or less frequent cycle of greening and drying out will not appear strongly in the dataset. This is due to the baseline image being quite dry. As a result this area will appear sparsely vegetated in the baseline and little change will be seen to occur.

  • Burns with low severity that only burn the understory in forested areas will be difficult to identify. This is due to the remaining green canopy obscuring the signal of the burn. This is likely to affect the detection and mapping of cool season control burns.

Quality assurance

Product accuracy is firstly dependent on the accuracy of the underpinning Earth Observation data. The daily Sentinel-2 (A and B combined) NRT provisional satellite data provides analysis-ready data that is processed on receipt using the best-available ancillary information at the time to provide atmospheric corrections.

These Earth Observation metrics are based on rigorous and peer reviewed research and have been used within the wider scientific community to understand burn characteristics within the landscape. However, this methodology has not been used as a near real-time and continuously processing continental scale output, such as in our web mapping service. The use of these metrics and this data is preliminary and provisional in nature and is still undergoing further development. These metrics should be used as a preliminary screening tool, and not an accurate identification of fire extent. These metrics should be used in combination with each other and can be used with other datasets to strengthen the agreement that the area has indeed been burnt. No decisions on life or property should be made based on this data.

To mitigate the impact of poor quality observations and other artefacts, locations with nodata (-999) values have been removed from the input dataset and then the data are screened using the Pixel Quality product (fmask) to remove cloud, cloud shadow, water and snow; leaving only “valid” data. However, it should be noted some poor pixels are likely to still be included in the data after screening.

Access the data

Get the data online

Learn how to access the data via AWS

Use constraints

DEA Burnt Area Vectors is a provisional dataset and should not be used for safety of life decisions.

Fires are fast moving and the situation on the ground can change rapidly. No decisions on life or property should be made based on this data. For local updates and alerts, please refer to your state emergency or fire service.

Other versions

You can find the history in the latest version of the product.