LiDAR-change-based mapping of sediment movement from an extreme rainfall event

ABSTRACT

Mapping landscape change at fine scales (e.g. <1.0 m resolution) using airborne LiDAR data from manned aircraft is a significant challenge. This challenge is magnified in disaster response contexts. A combination of collection and processing factors contributes to horizontal and vertical errors (and resulting uncertainty) in each pre- and post-LiDAR derived digital elevation model (DEM). Subsequently, the errors in the change surface from the two (or more) DEMs are an accumulation of the errors in the individual DEMs. Thus, reliable mapping erosion/deposition changes at sub-meter precision in change detection studies using LiDAR data is largely the domain of terrestrial LiDAR or sUAS with LiDAR scanners rather than manned aircraft. Unfortunately, terrestrial and sUAS LiDAR scanners are not well suited for mapping large areas and sUAS collections are subject to additional airspace constraints compared to manned aircraft. In this study, we probed one of the significant issues in airborne LiDAR change projects – vertical height errors from sequential flight lines. A simplified solution for determining flight line vertical biases in areas of low topographic relief with natural cover types was developed and tested for normalizing point clouds. The approach was tested in a fine-scale erosion/deposition study from an extreme rainfall event that eroded and deposited sand at depths of about 1.0 m. Airborne LiDAR had been collected prior to the rainfall event, and another airborne LiDAR collection was made 1 month after the event. Eleven field campaigns to collect reference data and visit anomalies in the change surface were conducted in a 15-month period after the event, beginning 25 February 2016 and ending 8 May 2017. The validation results indicate accuracies for the pre-event and post-event LiDAR derived DEMs were 7.8 cm and 13.0 cm RMSE, respectively. After modeling vertical errors and corrections applied to the post-event point clouds, the RMSE for the post-event DEM was 8.3 cm. In the depositional use case, 27 locations were sampled with auger boreholes/sand pits and compared with LiDAR-based change. The LiDAR-based change detection analysis resulted in predicted sand depth accuracies of 94% with a mean error of 4.7 cm.

KEYWORDS:

• LiDAR
• change detection
• flood
• sand deposition
• military training area

Acknowledgments

We express our appreciation to Chris Robbins at the South Carolina Army National Guard (SCARNG) and William Poulson and James Hibbert at the University of South Carolina. Special appreciation to Bruno Piovan who designed and crafted a custom sand auger, with an enclosed sleeve, for the stratigraphic field work.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The final DEMs, both pre and post- event, with the supporting ancillary data are available at: https://osf.io/7axbw/?view_only=e3fcb252d0d24761ba3f86eebb8f6543. https://osf.io/7axbw/?view_only=e3fcb252d0d24761ba3f86eebb8f6543.

The field data that support the findings of this study are available from the corresponding author, upon reasonable request. Requests for the airborne LiDAR point cloud must be made to the South Carolina Army National Guard.

Notes

1. We note the use of the terms digital elevation model (DEM) and digital terrain model (DTM) for describing a regular spaced raster of the bare-earth terrain surface in both the practice and literature. We use the acronym DEM exclusively, following USGS and the preponderance of literature.

Additional information

Funding

This research was carried out with the support of the University of South Carolina contract to SCARNG (#21600-FC56) This article has been published with funds provided by the PRIN project ”SYLVA”, (Prot. 20177H5KLM).