Wind erosion monitoring of moss crusts-covered slopes based on SfM photogrammetry technology

Wind erosion is a common natural disaster in arid and semiarid regions, particularly in desertified areas that are frequently subjected to droughts and harsh environment conditions. Biocrusts, widely distributed in these regions, play a significant role in controlling wind erosion. In recent years, the significance of biocrusts in controlling soil erosion and their role in enhancing the resilience of ecosystems in drylands have been increasingly acknowledged by researchers and practitioners in environmental science and land management. Quantitative monitoring and numerical modeling of wind erosion processes on slopes are crucial for analyzing the mechanisms of wind erosion, as well as predicting the effects of different types of biocrusts on erosion control. These analyses provide valuable data for environmental protection and land rehabilitation efforts, especially in arid regions where soil degradation is widespread. This study selected Structure-from-Motion photogrammetry technique to monitor morphological changes in moss crust-covered slopes after wind erosion, aiming to evaluate its accuracy and applicability. SfM photogrammetry is a cutting-edge method that uses multiple photographs taken from different angles to generate three-dimensional models of the landscape, allowing for precise monitoring of changes in surface morphology over time. In order to validate the rationality and accuracy of using this method to monitor wind erosion on moss crust-covered slopes, indoor wind tunnel experiments were conducted in wind tunnel laboratory at Institute of Soil and Water Conservation, Northwest A&F University. For this study, we collected samples from Dingbian Country, Shaanxi Province, China, where typical ecosystems are located. Soil samples and moss crust were collected from the topsoil of agricultural fields separately. Soil samples were carefully packed into erosion pans, which were placed in the wind tunnel for simulated wind erosion experiments. The erosion pans, designed with size of 1.0 m (length) × 0.6 m (width) × 0.2 m (height) , had known coordinate points distributed in a regular pattern around their perimeter, and the soil within them was arranged in layers to simulate natural soil stratification. Moss crust samples were transplanted piece by piece according to the different coverage percentage treatments, ensuring that the surface cover was consistent with the experimental conditions. The camera employed for SfM photogrammetry technique was a Canon EOS 90D single lens reflex (SLR) camera equipped with a zoom lens that covered a focal length range of 18 – 135 mm and the focal length was locked at 18 mm during the measurement to capture the details of the surface topography accurately. This study used the Agisoft PhotoScan Professional software to complete self-calibrating of camera, detecting and matching feature points, sparse reconstructing of point cloud, bundling adjustment and patch-based multi-view stereo, resulting in a dense point cloud and thereby a raster Digital Elevation Model with a pixel size of 0.3 × 0.3 mm, providing a highly detailed representation of the terrain. The estimated soil loss rates for the same erosion pan are based on three-dimensional model at the different stage. At the same time, the erosion was weighed using an accuracy weighing scale with a resolution of 0.001 kg in order to calculate the soil loss rates. Results indicated that the SfM Photogrammetry provided more than 80% accuracy in estimating wind erosion rates, with the absolute error ranges from 0.011 to 0.104, and the average relative error ranges (RME) from 2.91% to 18.15%. The root mean squared error (RMSE) from 0.015 to 0.105, demonstrating the effectiveness and reliability of this method. The analysis of the reconstructed point cloud from the SfM photogrammetry technique shows that the number of dense clouds ranged from 6,267,557 to 6,841,793 points. The distribution of these points was uniform, with densities ranging from 1,045 to 1,140 points per square centimeter, and the spacing between points was approximately 0.2 mm. There was no significant difference in the spacing between points for the various treatments, confirming the consistency of the data. Comparing the measured accuracy with the theoretical coordinate accuracy indicated that the error and relative accuracy were all within permissible limits, and the root mean squared error (RMSE) was 0.045 mm. The three-dimensional reconstruction model of the slopes provides an accurate description of the soil and moss crust patches, including the sections of erosion and deposition. The findings demonstrate that the SfM photogrammetric technique is an effective and dependable method for acquiring elevation data before and after the occurrence of wind erosion on moss crust-covered slopes, offering insights into the effectiveness of moss crusts in mitigating the impacts of wind erosion in arid and semiarid regions. This research contributes to our understanding of the role of biocrusts in ecosystem services and provides a valuable tool for land management practices aimed at combating desertification.