Camera-based soil surface change measurement is a cost-efficient and non-invasive approach to assess soil erosion. A challenging aspect in this context is the obscuring of the sediment yield by subsidence phenomenon such as soil consolidation and compaction in the beginning of a rainfall event (masking effect). Based on the camera elevation changes and measured field observations, we develop an approach to estimate these masking effects and to approximate a correction function. We therefore conduct ten rainfall simulations (3 m x 1 m) on different agricultural slopes, measuring runoff and sediment concentration. With a time-lapse camera system, we generate high resolution digital elevation models every 20 s. An s-shaped curve is fitted via non-linear regression for every rainfall simulation. We use the variables of these functions as well as a combination of the different field observations – bulk density, soil moisture, grain size distribution, total organic carbon, slope steepness, surface cover and surface roughness – as input values for an adjustment. We are able to estimate the masking effects at the beginning of rainfall events as functions of soil and plot characteristics and therefore offer a potential to increase the informative value of camera-based soil erosion measurements on agricultural fields.

中文翻译：

---

解码降雨对土壤表面变化的影响：延时 SfM 摄影测量中沉积物产量的经验分离


基于摄像头的土壤表面变化测量是一种经济高效且非侵入性的土壤侵蚀评估方法。在这种情况下，一个具有挑战性的方面是沉降现象（例如降雨事件开始时的土壤固结和压实）掩盖了沉积物的产量（掩盖效应）。根据相机仰角变化和测量的现场观察，我们开发了一种方法来估计这些掩蔽效应并近似校正函数。因此，我们在不同的农业斜坡上进行了 10 次降雨模拟 （3 m x 1 m），测量径流和沉积物浓度。借助延时摄影系统，我们每 20 秒生成一次高分辨率数字高程模型。通过非线性回归对每次降雨模拟进行 S 形曲线拟合。我们使用这些函数的变量以及不同现场观测值的组合——堆积密度、土壤水分、粒度分布、总有机碳、坡度陡度、表面覆盖和表面粗糙度——作为调整的输入值。我们能够将降雨事件开始时的掩蔽效应估计为土壤和小区特征的函数，因此有可能增加基于相机的农田土壤侵蚀测量的信息价值。