Introduction

Rill erosion is an important process of slope hydraulic erosion, being the main component of slope hydraulic erosion prediction models (Foster et al., 1995; Shen et al., 2015; Assouline et al., 2017; Zhang et al., 2019). Rill erosion not only accelerates slope erosion and the transport of erosion products (Guo et al., 2020; Gordon et al., 2012, Vereecken et al., 2016), it also controls the development of watershed geomorphology and the evolution of morphological characteristics (Mancilla et al., 2005; Oz et al., 2017; Wu et al., 2018). In rill erosion processes, rill morphology is the result of interactions between hydrodynamic factors and soil properties (Schneider et al., 2013; Assouline et al., 2017; Wu et al., 2020). Although the presence of rills results in more water flowing into rills from inter-rills, thereby increasing rill flow erosivity and the sediment transport capacity of runoff (Robichaud et al., 2010; Wu et al., 2018), an increase in runoff erosivity and sediment transport capacity aggravates rill morphology evolution (Schneider et al., 2013; Zhang et al., 2019). Therefore, the evolution of rill morphology, hydrodynamic characteristics and soil erosion form a complex mutual feedback process. Quantitative description of rill morphology and the relationship between rill morphological evolution, runoff and sediment yield are important to fully understand rill erosion, promoting the development of soil erosion simulation.
Currently, investigations have been undertaken on rill cross-section shape (rill width, depth, width depth ratio, etc.) (Mancilla et al., 2016; Shen et al., 2016; Oz et al., 2017) and rill network characteristics (average rill azimuth, rill density, rill cutting degree and average rill bending complexity, rill number and discontinuous rill number, rill fractal dimension) (Gómez et al., 2003). The development process of rills has been quantified from three perspectives (Govers et al., 2007; Shen et al., 2015; Zhang et al., 2017; Wu et al., 2018), as well as using the degree of connectivity (the number of combined nodes and bifurcation ratio) (Schneider et al., 2013; Zhang et al., 2017). Although these methods provide indices for characterizing rill networks, the indicators ignore rill evolution during rainfall runoff and soil erosion processes. In order to improve erosion model accuracy, these rill characteristics have recently been incorporated into slope erosion models, resulting in an improvement in model results. For example, Wu et al. (2020) constructed a model to embed rill evolution module into hillside rainfall runoff erosion processes. However, their experimental materials were mainly concentrated in one or a few types of soil, and the effects of soil properties on rill erosion morphology and sediment yield were not systematically considered. Due to changes in soil properties and the complexity of soil erosion processes, each function can only be used for the soil erosion model under similar soil or experimental conditions.
The effects of soil properties on slope erosion and rill morphology are often ignored. Soil type and soil texture have significant effects on soil separability, particle size distribution of eroded sediment and morphological characteristics of erosion (Knapen et al., 2007). The difference of soil physical properties is the internal influencing factor of erosion processes and rill erosion (Dong et al., 2014; He et al., 2014; Wu et al., 2017). Soil properties determine the soil shear strength and stripping rate, directly affecting the occurrence of slope erosion. Bonilla et al. (2012) found that the correlation between silt content and soil separability was the largest, followed by sand content, while the correlation between clay content and soil erodibility was the weakest. Processes of soil separation and sediment transport are also closely related to the size of primary and aggregate soil particles (Rienzi et al., 2013). At the same time, soil properties affect the formation and stability of soil crusts, change surface soil permeability, and regulate runoff and runoff velocity, having an important impact on slope erosion and rill morphology (Vaezi et al., 2017; Mahalder et al., 2018; Sun et al., 2020). Soil properties are also closely related to soil structure stability. Under different rainfall intensities, different soil structure stabilities will have an important impact on gully wall stability (Xu et al., 2015; Guo et al., 2019).
As a key agricultural area in northwest China, the Loess Plateau has suffered from serious soil erosion for decades, and rill landforms formed by water erosion have serious impacted agricultural production (Wu et al., 2018; Sun et al., 2020). At the same time, soil in the Loess Plateau has a consistent particle size composition (0.05-0.005 mm silt, accounting for 58% - 75%) and zonality of particle size distribution. Zonality shows that grain size of loess gradually becomes fine from the northwest to the southeast, coinciding with sandy soil, light soil, medium soil and heavy soil zones, respectively. Loess grain size differences in the north-south direction have a profound impact on landforms and soil erosion. Although many experiments have been undertaken to study the process and mechanism of water erosion in the Loess Plateau (Fang et al., 2015; Shen et al., 2016; He et al., 2017), it is still unclear how changes to soil properties in different regions affect rill morphology during surface erosion, and the accuracy of the regional erosion prediction model with multiple soil types needs to be further improved. Therefore, a series of simulated rainfall experiments were carried out in different soil zones to examine the following objectives: (1) to examine the effects of soil characteristics on rill morphology (2); to discuss differences in different types of soil slope erosion responses to rainfall intensity and slope angle; and (3) to establish a prediction model of slope erosion based on soil characteristic factors and rill shape factors. Findings from this investigation will improve our understanding of erosion mechanisms of multiple soil zones.