The effect of land uses and rainfall regimes on runoff and soil erosion in the semi-arid loess hilly area, China

Summary

The main purpose of this article is to analyze runoff and soil loss in relation to land use and rainfall regimes in a loess hilly area of China. Based on 14 years of field measurements and K-means clustering, 131 rainfall events were classified into three rainfall regimes. Rainfall Regime II is an aggregation of rainfall events with such features as high intensity, short duration and high frequency. Regime I is the aggregation of rainfall events of medium intensity, medium duration and less frequent occurrence. Regime III is the aggregation of events of low intensity and long duration and infrequent occurrence. The following results were found. (1) Mean runoff coefficient and erosion modulus among the five land use types are: cropland > pastureland > woodland > grassland > shrubland. (2) The sensitivity of runoff and erosion to the rainfall regimes differ. Rainfall Regime II causes the greatest proportion of runoff and soil loss, followed by Regime I and Regime III. (3) The processes of runoff and soil loss, however, are complicated and uncertain with the interaction of rainfall and land use. This is mainly due to the different stages of vegetation succession. Based on these results, it was suggested that more attention should be paid to Rainfall Regime II since it had the most erosive effect. Shrubland is the first choice to control soil erosion when land use conversion is implemented, whereas pastureland (alfalfa) is not. Large-scale plantation of alfalfa therefore, should be avoided. Grassland and woodland can be used as important supplements to shrubland.

Introduction

Soil erosion, defined as the detachment and displacement of soil particles from the surface to another location (Govers et al., 1990, Flanagan, 2002), continues to be a primary cause of soil degradation throughout the world (Fu and Gulinck, 1994), and has become an issue of significant and severe societal and environmental concern (Elsen et al., 2003, Singha et al., 2006). Land use/cover, as one of the most important factors, influences the occurrence and the intensity of runoff and sediment yield (Hovius, 1998, Karvonen et al., 1999, Chen et al., 2001). Non-uniform variations in land use/vegetation coverage proved to be closely related to hydrological responses over catchments (Siriwardena et al., 2006). By properly adjusting of land use/land cover patterns, soil properties can be greatly improved, consequently reducing soil erosion to the allowed threshold (Fu, 1989, Chen et al., 2003), and the improved soil physical properties can also positively affect the establishment of vegetation (Kosmas et al., 2000). On the other hand, improper land use and/or cover patterns can cause severe water, soil and nutrient losses, and further land degradation (Luk et al., 1989, Costa et al., 2003).

Runoff and erosion processes, however, are strongly affected by many other factors besides land use/land cover. Among these factors, the one most mentioned is rainfall. Rainfall can cause soil erosion and runoff when it reaches the ground (Sharma et al., 1993, Dijk et al., 2002, Kinnell, 2005). Also, the spatiotemporal heterogeneity and uneven characteristics of rainfall play a key role in soil erosion (Li et al., 2000, Nearing, 2001, Bürger, 2002, Endale et al., 2006). Morin et al. (2006) found that complex interactions exist between the spatiotemporal distributions of rainfall systems and watershed hydrological responses. Local storm patterns are important in determining the shape of the runoff hydrograph (de Lima and Singh, 2002). Runoff and sediment generation in different land use types may thus vary greatly with various rainfall types. Addressing the response of runoff/erosion to different land use/land cover types and different rainfall types is therefore important for land use structure adjustment and vegetation restoration.

Rainfall classification, however, is an important problem, which needs to be solved. Most studies focused on the response of the runoff/erosion process to single rainfall pattern and different vegetation types (Yeh et al., 2000, de Lima et al., 2003, Kirkby et al., 2005). Undoubtedly, controlling soil erosion requires much more detailed and accurate data in the real world (Elsen et al., 2003). Many studies, however, are based on rainfall simulations, and thus the conclusions are often not applicable to the real world. For example, some authors have suggested that the nozzles of rainfall simulators produce low kinetic energies relative to natural rainfall (Luk et al., 1986). Madden et al. (1998) also found that the kinetic energy and erosivity of rainfall produced by simulators could be lower than that of natural rainfall. This insufficiency of energy plays an important role in infiltration capacity, preventing surface crusting and sediment detachment (Mathys et al., 2005). Other results also show that rainfall simulators are unable to reproduce natural rainfall conditions (Aizen et al., 2000, Mazi et al., 2004, Nearing et al., 2005). Accordingly, finding real rainfall-runoff-sediment patterns based on measurements are important for soil erosion control.

Soil loss and runoff studies at plot scales have been confirmed to be of crucial importance (Licznar and Nearing, 2003). Reliable and consistent erosion measurements and extensive field data have played primary roles in soil erosion analysis and prediction on larger scales (Zhang et al., 1996, Nearing et al., 1999). In addition, soil properties are always affected by land uses/vegetation evolution over long time scales (e.g., months-centuries), which then further influence runoff and soil erosion (Eagleson, 1982, Xu, 2005). For example, the accumulation of litter under plants contributes to increased surface roughness, higher infiltration rates, and decreased runoff generation thresholds (Boer and Puigdefábregas, 2005). Bochet et al. (1999) also found that topsoil modification and erosion processes are mainly due to the differential influences of species morphology (i.e., above-ground structure) and components (i.e., litter cover and organic matter). Moreover, the impact of plant roots on soil resistance to erosion by water is also significant (Hou, 1990, Zou et al., 2000, Gyssels et al., 2005, Mao et al., 2006). In general however, these kinds of long-term consecutive studies in arid and semi-arid areas are relatively scarce.

In this study, based on 14 years of field measurements in plots in a semi-arid loess hilly area, 131 rainfall events that produced runoff were recorded. On the basis of rainfall depth, duration and maximum 30-min intensity, all the events were classified into three categories. They were then used to analyze the effects of varying land uses and rainfall regimes on runoff and soil erosion. The specific objectives were: (1) to analyze the effects of land use/land cover on soil and water loss, (2) to determine the response of runoff and soil erosion to different rainfall regimes, and (3) to study the role of different land use types on soil erosion control under different rainfall regimes.