Annual vegetation height for Australia for 2019 to 2024

Full description

This product provides an annual estimate of vegetation height (top of canopy) for 2019 to 2024 at a 30m pixel size. These images are created using an extreme gradient boosting regressor (XGBoost) machine learning model which was training using GEDI vegetation height (from the 98th percentile height) with spatially complete earth observation data as independent variables: Landsat annual surface reflectance, Landsat annual fractional cover percentiles, PALSAR backscatter (HV polarised) annual mosaic, along with a Digital Elevation Model (DEM) as well as long-term average temperature and rainfall. Further details are provided in (Ticehurst et al. (in review)). This product was created as part of the National Bushfire Intelligence Capability (NBIC) and supported by the Terrestrial Ecosystem Research Network (TERN).

Ticehurst C, Joshi R, Hussain S, Walker S, Opie K, Donohue R (in review) Developing a nation-wide vegetation height layer for bushfire fuel classification. Submitted to ISPRS Journal of Photogrammetry and Remote Sensing.
Lineage: GEDI data are available on NASA’s Earth Data search website (https://search.earthdata.nasa.gov/). All available GEDI L2A data for Australia from 2019 to 2023 were downloaded and cleaned using the standard quality parameters. These cleaned data were used to train the model. The Landsat annual surface reflectance and fractional cover percentiles are generated through Digital Earth Australia (DEA; https://www.ga.gov.au/dea/home). These data have been indexed in the DEA datacube and are available in the CSIRO EASI platform (https://research.csiro.au/easi/). The annual PALSAR backscatter mosaics are provided by the Japanese Aerospace Exploration Agency (https://www.eorc.jaxa.jp/ALOS/en/dataset/fnf_e.htm). The DEM was derived from the Shuttle Radar Topography Mission (SRTM) data (DEM-H) and is also available in the CSIRO EASI platform. The long-term average rainfall and temperature data were available from the Bureau of Meteorology. A permanent water mask, which was applied to the annual vegetation height mosaics, was available from https://glad.umd.edu/dataset/gedi/. The vegetation height product was generated using Jupyter notebooks on the CSIRO EASI platform.
Further details about product lineage are provided in the product description pdf file (AnnualVegetationHeight_ProductDescription.pdf) located under Supporting Files.