UAV-SfM 4D mapping of landslides activated in a steep terraced agricultural area

Abstract

The presence of roads is closely linked with the activation of land degradative phenomena such as landslides. Factors such as ineffective road management and design, local rainfall regimes, and specific geomorphological elements actively influence landslide occurrence. In this context, recent developments in digital photogrammetry (e.g., Structure from Motion; SfM) paired with Uncrewed Aerial Vehicles (UAV) systems increase our possibilities to realize low-cost and recurrent topographic surveys. This can lead to the development of multi-temporal (hereafter: 4D) and high-resolution Digital Elevation Models (DEMs), which are fundamental to analyse geomorphological features and quantify processes at the fine spatial and temporal resolutions at which they occur. This research proposes a multi-temporal comparison of the main geomorphometric indicators describing a landslide-prone terraced vineyard to assess the observed high-steep slope failures. The possibility to investigate the evolution of landslide geomorphic features in steep agricultural systems through a high-resolution and 4D comparison of such indicators is still a challenge to be explored. In this article, we considered a case study located in the central Italian Alps, where two landslides were activated below a rural road within a terraced agricultural system. The dynamics of the landslides were monitored by comparing repeated DEMs (DEM of difference), which reported erosion values of above 20 m3 and 10 m3 for the two landslide zones and deposition values of more than 15 m3 and 9 m3, respectively. The road network’s role in the alteration of superficial water flows was proved by the elaboration of the relative path impact index. Altered water flows were expressed by values between 2σ and 4σ close to the collapsed surfaces. The increase in profile curvature and roughness index described the landslides evolution over time. Finally, the multi-temporal comparison of feature extraction underlined the geomorphological changes affecting the study area. The accuracy of features extraction was analysed through the quality index computation, expressed in a range between 0 (low accuracy) and 1 (high accuracy), and proved to be equal to 0.22 m (L1-pre), 0.63 m (L1-post), and 0.69 m (L2). Results confirmed the usefulness of high-resolution and 4D UAV-based SfM surveys to investigate landslides triggering due to the presence of roads at hillslope scale in agricultural systems. This work could be a useful starting point for further studies of landslide- susceptible zones on a wider scale to preserve the quality and the productivity of affected agricultural areas.

Soil and water conservation in terraced and non‐terraced cultivations: an extensive comparison of 50 vineyards

“…Furthermore, the introduction of machinery has implied the need for access roads, which typically represent the steepest and most hydrologically connected parts of the vineyard (Mauri et al, 2021a; Pijl et al, 2019a). Hillslope roads are a widely recognised SWC challenge related to anthropogenic and, particularly, cultivated hillslopes (Preti et al, 2018b; Tarolli et al, 2015), producing high rates of erosion (Pijl et al, 2019a; Sidle & Ziegler, 2012) and potentially contributing to landslide initiation (Galve et al, 2016; Mauri et al, 2021b; Zingaro et al, 2019).…”

Review on the Geophysical and UAV-Based Methods Applied to Landslides

“…Numerous SfM methods now address the determination of the 3-D location of matching features in multiple photographs taken from different angles to identify features in individual images and find tie points [87]. UAV-SfM 4D mapping of landslides is also possible with a multi-temporal comparison of geomorphometric indicators and feature extraction [88]. As shown in Figure 3, adaptive sensors are designed for UAVs for hyperspectral imaging, LiDAR, synthetic aperture radar and thermal infrared units [6].…”

Assessing Many Image Processing Products Retrieved from Sentinel-2 Data to Monitor Shallow Landslides in Agricultural Environments

“…Recently, experiments have employed Synthetic Aperture Radar (SAR) images [7,8] although detecting and mapping landslides using SAR remains challenging [9]. Additionally, numerous studies in agricultural areas have employed an Unmanned Aerial Vehicle (UAV) [10][11][12][13].…”