Spatio-temporal monitoring of piping and assessment of erosion and sedimentation using Multi-temporal UAV data

Extended Abstract
Introduction
Soil erosion is one of the most critical environmental challenges in semi-arid regions worldwide, particularly in landscapes dominated by loessic deposits, where weak physical and mechanical characteristics significantly increase susceptibility to subsurface erosion and the development of piping features. Piping often initiates and progresses covertly during its early stages, eventually leading to sudden surface collapse, topographic instability, accelerated sediment delivery, reduced land productivity, and disruption of hydrological and ecological systems. Despite extensive research on soil erosion, most existing studies have adopted static or single-temporal approaches, and substantial scientific gaps still remain regarding multi-temporal and spatial monitoring of piping frequency, distribution, density, and evolutionary trends. Furthermore, the combined and interactive influences of topography, vegetation cover, and land use patterns on piping development in loess-derived terrains are not yet adequately understood, posing challenges for soil and water conservation planning, bioengineering practices, and watershed management decisions. Accordingly, the present study aims to monitor four-year temporal changes in the number, location, and evolution of piping features, and to evaluate the controlling roles of slope, elevation, vegetation cover, and land use using ultra-high-resolution UAV data combined with multi-temporal Digital Elevation Model of Difference (DoD) analysis. This approach enables the assessment of spatial patterns of erosion and deposition in areas with and without piping and the estimation of annual erosion-sedimentation rates, thereby improving the identification of high-risk zones and supporting evidence-based management and mitigation strategies in loess environments.
Materials and Methods
Initially, two sub-watersheds with different proportions of rangeland and agricultural land use were selected. To ensure accurate detection and temporal monitoring of piping development, the spatial location of all piping features within both sub-watersheds was recorded using GPS during the 2019 and 2023 survey campaigns. Multi-temporal UAV surveys were conducted under comparable illumination and meteorological conditions using a Phantom 4Pro UAV, and the acquired high-resolution imagery was processed using a photogrammetric workflow in ContextCapture to generate three-dimensional point clouds and high-precision Digital Elevation Models (DEMs) with a spatial accuracy of approximately 5 cm. To quantify volumetric topographic changes, the DoD approach was applied within ArcGIS, resulting in spatially explicit erosion deposition maps as well as annual mean volumetric change estimates (expressed as tons per hectare per year) for each sub-watershed. Land use classification was carried out through visual interpretation of UAV imagery combined with extensive field verification. Slope and elevation layers were extracted from the DEM using ArcGIS to examine topographic control on piping distribution. Density plots generated in R software were used to statistically explore the relationships between piping occurrence, slope gradient, and elevation range. Finally, temporal variations in piping frequency, spatial displacement, initiation, expansion, or disappearance were compared between the two sub-watersheds to identify dominant geomorphic and land-management drivers of piping dynamics.
Results and Discussion
The findings indicated that in Sub-watershed 1, with 85% agricultural land, the piping density was only 10%, of which 2 occurred in croplands and 18 in rangelands. In Sub-watershed 2, with 70% rangeland, the density was considerably higher at 55%, with 198 cases occurring in rangelands. Piping mainly occurred at lower elevations (370–410 m and 300–340 m in Sub-watersheds 1 and 2), on steep slopes (25–35°), and weak vegetation. DoD analysis over the period 2019–2023 revealed that in agricultural lands, deposition was the dominant process, whereas in rangelands, erosion  were more pronounced; in Sub-watershed 1, 72% of the area experienced deposition and 28% erosion, while in Sub-watershed 2, 77% erosion and 23% deposition were recorded. Annual rates were ±5 t/ha/yr in Sub-watershed 1 and 15–25 t/ha/yr in Sub-watershed 2. Over four years, two agricultural piping features were lost, but two new features formed in Sub-watershed 1 and ten in Sub-watershed 2. The main advantage of this study lies in the integration of real UAV data, precise DoD analysis, pixel-based monitoring of erosion, and piping relocation, enabling identification of high-risk areas and prioritization for management interventions.
Conclusion
Based on the findings of this research, Steeper slopes, lower elevations, reduced vegetation density and land-use type were identified as the primary environmental factors controlling the initiation and development of piping in semi-arid loess landscapes. Moreover, the integration of multi-temporal UAV data with the DoD technique enabled accurate detection of morphological evolution and delineation of susceptible areas over time. According to the results, Sub-watershed 1, dominated by agricultural land use, was mainly characterized by depositional processes, and the total number of piping features remained constant during the four-year monitoring period. In contrast, Sub-watershed 2, where rangelands are dominant, experienced severe erosion, resulting in the formation of eight new piping. This discrepancy can be attributed to contrasting land management practices: agricultural operations such as tillage and crop cultivation may lead to the infilling or concealment of existing pipes, whereas terrain forms, overgrazing and vegetation degradation in rangelands facilitate accelerated piping expansion. Field observations also revealed a dual functional role of piping features. While they intensify subsurface discharge, soil erosion, and desertification, their internal cavities may also serve as favorable microhabitats for the establishment of drought-resistant plant species such as wild pomegranate and Amygdalus scoparia. The outcomes of this research can directly support soil and water conservation planning, particularly for prioritizing preventive measures in fragile dryland environments. Enhancing deep-rooted vegetation, regulating grazing patterns, and applying bio-engineering strategies around piping zones are recommended for controlling further degradation. Future studies are advised to integrate UAV observations with seasonal satellite datasets and high-resolution DEM modeling while assessing climate-change-driven rainfall scenarios to better predict long-term piping dynamics.