Geomorphic change detection between 2018 and 2024 for the Fitzroy River Basin Queensland Australia

Full description

This dataset shows geomorphic change in the Fitzroy River Basin Queensland, Australia. The change is determined by the difference in elevation in the overlapping areas of two co-registered airborne laser scanning (ALS) point cloud datasets, collected in 2018 and 2024.

The change is measured in metres, with negative values indicating erosion and positive values indicating deposition.
Lineage: INPUT DATASETS
Airborne laser scanning (ALS) datasets
- Fitzroy 2018 capture (Australian Government Department of Climate Change, Energy, the Environment and Water)
- Fitzroy 2024 capture (Australian Government Department of Climate Change, Energy, the Environment and Water)
These ALS datasets are publicly available from the Geoscience Australia Elevation and Depth Spatial Data Portal https://elevation.fsdf.org.au/

METHOD
The method used to create these geomorphic change detection (GCD) datasets is from: Walker et al. 2025. Optimising sub-metre resolution 3D geomorphic change detection over large areas using multitemporal airborne laser scanning with Sentinel-1 InSAR and Sentinel-2 optical observations. Remote Sensing of Environment 317. https://doi.org/10.1016/j.rse.2024.114522
This method used the following steps:
1.\tCo-registration of the 2018 and 2024 ALS point cloud datasets by aligning the two elevation point clouds in the x, y and z dimensions. Sentinel-1 and Sentinel-2 data were not used to determine suitable areas in the co-registration process (as in Walker et al., 2025). Instead, the point clouds were aligned by using powerline and power station infrastructure across the study area.
2.\tAll lidar points in each point cloud were then assigned to corresponding ‘lidar cells’ which are 0.5 m x 0.5 m grid cells containing all lidar points within the cell’s coordinate bounds.
3.\tCalculation of an initial 3D GCD estimate as the distance between 2018 and 2024 ground surface, where the ground surface elevation is determined by the median distance of points from the best-fit plane in each lidar cell. Inputs are the co-registered ALS points in lidar cells from Step 1.
4.\tCreation of a land surface classification using a supervised random forest model, with the following inputs combined as a three-band composite image; (i) an initial 3D GCD estimate from Step 3, (ii) ALS intensity data, and (iii) within-cell elevation standard deviation.
5.\tEstimation of error from vegetation-related point misclassification determined from the land surface classification from Step 4.
6.\tCalculation of spatially-variable limit of detection (of change) using detrended elevation standard deviation in the lidar cells from the 2018 and 2024 ALS point clouds from Step 2.
7.\tAdjusting the limit of detection to incorporate the estimated vegetation-related error (Step 5).
8.\tManually-created mask of surface changes caused by human activities, such as mining, dam excavation, road works, and erosion remediation, to remove these sources from GCD values and summary totals.
9.\tFinal 3D GCD estimate per grid cell combining; (i) initial 3D GCD estimate, (ii) adjusted limit of detection from Step 6, and (iii) mask of surface changes from Step 7.

OUTPUT DATASETS
- GCD tiles (GeoTIFF) with full uncertainty analysis (Steps 4, 5, 6, 7 and 8) showing elevation change in each 0.5 m x 0.5 m grid cell (units in metres). Tiles are named according to the geographic coordinates (GDA2020 Zone 56) of the lower left corner (e.g., 182000_7407000_gcd_fitzroy.tif). These tiles contain geomorphic change estimated with very high certainty (95% confidence).
- GCD tiles computed using only Steps 1, 2 and 3 (from above) and with no uncertainty analysis applied (‘raw’ GCD). These tiles contain geomorphic change estimated with lower certainty (no uncertainty analysis).
- Google Earth .kmz files showing 95% confidence GCD results for all tiles in a single file and ‘raw’ GCD results for all tiles in a single file. Elevation changes in units of metres.
- A .shp GIS file for visualising the tile map.
- Tables (.csv) summarising the total erosion and deposition across each tile. For each tile the .csv includes (i) the tile coordinates, (ii) the area of the tile, (iii) the annual volume of deposition (iv) the annual volume of erosion, (v) the percentage area of the tile with registered deposition, and (vi) the percentage area of the tile with registered erosion.