2019-20 NSW East Coast Bushfire Season Analysis
As part of the ProRaster Scientific and ProRaster Bushfire development, I have been studying the 2019-2020 bushfire event on the East Coast of New South Wales, Australia. I have used Landsat 8 and Landsat 9 data. The goal was to test and refine the technology in ProRaster rather than draw any serious conclusions regarding the fire event.
It is useful to understand the following terms:
A scene is a single raster data acquisition event recorded by a satellite. For Landsat and Sentinel 2, the scene is the basic unit of data that you will download for analysis. Each satellite will revisit each scene periodically. The data for each scene acquisition is collocated, but the extent of coverage may vary slightly over time.
A temporal sequence of scenes. You can think of a sequence as a movie where each frame is a scene that was acquired at a known time.
A scene collation is a sequence of scenes that have been collated, or collapsed, to a single raster dataset. The scenes will be chosen to be as close to each other temporally as possible. The goal is to acquire high-quality pixels throughout the scene, removing pixels that are compromised by cloud or other issues.
A mosaic is one or more spatially separate, but adjacent scenes combined into a single raster dataset. A mosaic will provide greater spatial coverage than a single scene.
We can combine these concepts to generate a list of possible products. These include:
- Scene Sequence
- Collated Scene
- Collated Scene Sequence
- Scene Mosaic
- Scene Mosaic Sequence
- Collated Scene Mosaic Sequence
The data used in the fire study is an example of the last product – a “Collated Scene Mosaic Sequence”. I have looked at four adjacent scenes which cover the southeast coast of NSW. For each scene, I downloaded one or more acquisitions in close temporal proximity, and these were collated to improve data quality and coverage. Then I did this for each of the five years (2018 to 2022) to generate a collated scene mosaic sequence.
The report is available for download in PDF format here.