Give Me the Dirt: Detection of Gully Extent and Volume Using High-Resolution Lidar
The Fitzroy catchment drains into the Great Barrier Reef lagoon. It is the largest catchment on the east coast of Australia, with an area of over 140,000 km2. To ascertain point-sources of erosion, and to quantify the volume of sediment lost from gullies, within the Fitzroy catchment is a major challenge for land-management; despite this, gully locations and volumes have never been thoroughly investigated. This study aims to develop a semi-automated method to detect and map gully extent and volume, using aircraft-mounted Light Detection and Ranging (LiDAR) technology within the Fitzroy catchment. Twenty LiDAR transects were acquired in 2007 (5000 × 275 m). The average distance between points of the LiDAR data was 0.3 m on the ground, with a height accuracy of within 0.1 m. Digital Elevation Models were derived for the transects, with a 0.5-m spatial resolution. We delineated gullies using terrain attributes and the backscatter intensity of the LiDAR returns. Transects were classified as ‘gully’ or ‘non-gully’ using objected-oriented classification. Gully volume was estimated for each pixel of the twenty transects. For four transects, we used a random forest algorithm to model the relation between gully presence and a set of readily available ancillary variables. We also modelled the relation between gully volume and the ancillary variables. These models were used to predict gully presence and volume at unsampled locations. We considered the extrapolation a success. The products generated from this study will be used to inform water-quality models, to asses land condition, and to improve our understanding of the dynamics of gully erosion under different climate and land-management regimes.
KeywordsRandom Forest LiDAR Data Backscatter Intensity Unsampled Location Random Forest Classification
We would like to thank Dr Robert Denham for his statistical and programming advice.
- Armston JA, Danaher TJ, Collett LJ (2004) A regression approach for mapping woody foliage projected cover in Queensland with Landsat data. In: Proceedings of the 12th Australasian Remote Sensing and Photogrammetry Conference, Fremantle, Australia, October 2004.Google Scholar
- Breiman L (2001) Random forests. Machine Learning 45:5–32.Google Scholar
- Brough DM, Claridge J, Grundy MJ (2006) Soil and Landscape Attributes: A Report on the Creation of a Soil and Landscape Information System for Queensland. Natural Resources, Mines & Water, Brisbane, Australia. QNRM06186.Google Scholar
- Congalton RG, Green K (1999) Assessing the Accuracy of Remotely Sensed Data: Principles and Practices. Lewis Publishers, Boca Raton, FL.Google Scholar
- Definiens, (2006) Definiens Professional 5: User Guide. Document version 18.104.22.168, Definiens AG, Munich, Germany.Google Scholar
- Dougall C, Packett R, Carroll C (2005) Application of the SedNet model in partnership with the Fitzroy Basin community. In: Zerger A and Argent RM (eds.) MODSIM 2005 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, December 2005, pp. 1119–1125Google Scholar
- Dougall C, Packett R, Carroll C, Sherman BS, Read A, Chen Y, Brodie J (2006) Sediment and nutrient modelling in the Fitzroy NRM region. Volume 5 In: Cogle AL, Carroll C and Sherman BS (eds.) The Use of SedNet and ANNEX Models to Guide GBR Catchment Sediment and Nutrient Target Setting. Department of Natural Resources, Mines and Water, Brisbane, pp. 1–27.Google Scholar
- Dougall C, Carroll C, Herring M, Trevithick R (2007) Sednet modelling in the Fitzroy Basin 2007; spatially variable ground cover and revised gully layers can potentially generate significant changes in erosion sources and patterns. In: Oxley L and Kulasiri D (eds.) MODSIM 2007, International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, December 2007, pp. 881–887.Google Scholar
- Hughes AO, Prosser IP, Stevenson J, Scott A, Lu H, Gallant J, Moran CJ (2001) Gully Erosion Mapping for the National Land and Water Resources Audit, CSIRO Land and Water Technical Report, 26/01, August 2001.Google Scholar
- Joo M, Yu B, Carroll C, Fentie B (2005) Estimating and modelling suspended sediment loads using rating curves in the Fitzroy River catchment Australia. In: Zerger A. and Argent R.M. (eds) MODSIM 2005 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand December 2005 pp. 1161–1167.Google Scholar
- Lane SN, Westaway RM, Hicks M (2003) Estimation of erosion and deposition volumes in a large, gravel-bed, braided river using synoptic remote sensing. Earth Surface Processes and Landforms 28(3):249–271.Google Scholar
- Liaw A, Wiener M (2002) Classification and regression by random-Forest. R News 2(3). pp. 18–22.Google Scholar
- R Development Core Team (2008) R: A Language and Environment for Statistical Computing. (Vienna: R Foundation for Statistical Computing. ISBN 3-900051-07-0. URL: http://www.R-project.org; last accessed 13.6.2008).
- Scarth P, Byrne M, Danaher T, Henry B, Hassett R, Carter J, Timmers P (2006) State of the paddock: monitoring condition and trend in groundcover across Queensland. In: Proceedings of the 13th Australasian Remote Sensing Conference, November 2006, Canberra.Google Scholar
- Trevithick R, Herring M, Dougall C (2008) Development of gully length density layer for the Fitzroy Basin, Queensland through gully mapping and extrapolation using Cubist. Proceedings of the 14th Australasian Remote Sensing and Photogrammetry Conference, Darwin, 29th September–3rd October, 2008.Google Scholar