Measuring surface roughness of agricultural soils: Measurement error evaluation and random components separation

•Real DEM of six soil surfaces with different roughness was generated by LiDAR technology.•The determination method of roughness measurement accuracy was given out.•The anisotropy and multi-scale characteristics of soil surface roughness were analyzed.•High pass filtering method can extract random components of roughness. Soil surface roughness (SSR) is an important parameter for predicting soil erosion, modeling reflectivity in the optical and microwave bands, and estimating soil infiltration capacity. Influenced by the interactions of soil properties, human activities, and natural factors, the anisotropy and scale dependence of SSR make it difficult to obtain high precision measurement results. To quantify its measurement error and investigate its scale characteristics, six selected soil surfaces, each approximately 30 m × 30 m, with different roughness were scanned by the LiDAR (Light Detect and Ranging) technology. Their digital elevation models (DEM) at 1 cm spatial resolution was generated from scanned point cloud data. According to the random sampling theory and the spatial filtering method, the following results were obtained: 1) The root mean squared height (RMSH) of the six surfaces ranged between 2.08 and 5.2 cm, and an exponential function was more suitable to represent the spatial correlation of natural farmland surface; 2) SSR measurement error demonstrated 6 m segment length (L) and 20 repeated observations could obtain 80% accuracy for ridge structure soil surface. Non-ridge structure surfaces need shorter L (3 m) for the same SSR measurement accuracy compared with ridge structure surfaces; 3) The filter window size determination method based on correlation length (CL) could automatically estimate the random component of SSR, and a weakened anisotropy for the filtered SSR was found after filtering. The results of this study are beneficial to the determination of SSR measurement accuracy and the application of SSR in agriculture and remote sensing fields.