Assessing the repeatability of terrestrial laser scanning for monitoring gully topography: A case study from Aratula, Queensland, Australia

Abstract

Terrestrial laser scanning (TLS) technology is a powerful tool for quantifying gully morphology and monitoring change over time. This is due to the high sampling density, sub-centimetre positional accuracies (x, y, z), flexibility of survey configurations and ability to link multiple TLS scans together. However, to ensure correct interpretation of results, research is needed to test the repeatability of TLS derived products to quantify the accuracy and separate ‘false’ from ‘true’ geomorphic change. In this study, we use the RIEGL VZ400 scanner to test the repeatability of TLS datasets for mapping gully morphology. We then quantify change following a rainfall event of approximately 100mm. Our study site, located in south-east Queensland, Australia was chosen to be challenging from a repeatability perspective with high topographic variability. The TLS data capture involved three sets of linked scans: one survey pre-rainfall, to be compared to two surveys post-rainfall acquired on consecutive days. Change is considered negligible in the two post-rainfall scans to test survey repeatability. To verify TLS accuracy, an independent dataset of gully extent and spot heights were acquired using traditional total station techniques. Results confirm that the TLS datasets can be registered multi-temporally at sub-centimetre levels of accuracy in three dimensions. Total station and TLS elevation samples showed strong agreement with a mean error and standard deviation (SD) of residuals equal to 0.052 and 0.047m, respectively (n=889). Significantly, our repeatability tests found that return type and pulse deviation influence the accuracy and repeatability of DEMs in gully environments. Analysis of consecutive day datasets showed that DEMs derived from first return data recorded 40% higher SD of residual error than DEMs using multiple return data. A significant empirical relationship between pulse deviation and the variance of residuals for repeat DEMs is also shown (r2=0.95). By accounting for pulse deviation, we demonstrate that a spatially variable change detection threshold increased the geomorphic change detected following the rainfall event. Overall, this work supports TLS as a tool for monitoring change in gully morphology; however, consideration of the survey configuration is needed to maximise the change detected and ensure repeatability of results.