A procedure for assessing the impacts of land-cover change on soil erosion at basin scale

Soil erosion is one of the most serious and global ecological environmental crises in progress today (Renschler et al. 1999; Natalia et al. 2005). The problems that are caused by soil erosion include decreased soil fertility, increased landslide activity, reservoir sedimentation, contaminant diffusion, rocky desertification, and ecosystem disturbances, all of which significantly impact human development (Syvitski et al. 2005; Jiang et al. 2014). The assessment of soil erosion is critical, considering its significant impact on the ecological environment (Xin et al. 2011). Many methods have been attempted to estimate soil erosion situations. Among these methods, soil erosion models, including empirical models and physical models, are straightforward but effective tools. The first widely used empirical model, Universal Soil Loss Equation (USLE), was developed in 1965 based on numerous observational data and rainfall simulation experiments (Wischmeier & Smith 1965). Limited by some deficiencies, USLE was continuously revised, leading to a new version named Revised Universal Soil Loss Equation (RUSLE) in the 1990s (Renard & Ferreira 1992; Renard et al. 1997). This new model can expediently evaluate the average annual soil erosion amount under different land-cover conditions and predict the spatial heterogeneity of soil erosion, providing a basis for soil conservation planning (Tang et al. 2015). Although some physical models, such as the Water Erosion Prediction Project (Foster & Lane 1987), the European Soil Erosion Model (Morgan et al. 1998), the Limburg Soil Erosion Model (De Roo et al. 1996), and the Soil Erosion Model for Mediterranean regions (De Jong et al. 1996) have been proposed in recent decades, their applications are presently limited due to the requirement of many parameters and complex calculation process. Moreover, some data (e.g., temporally continuous rainfall data) are not available for many countries and regions, making them difficult to promote. Benefitting from the convenience in application and compatibility with geographical information systems (GIS), the USLE and RUSLE models are still the most frequently used soil erosion models worldwide.

In this study, an integrated approach linking the RUSLE model with GIS tools was used to assess the soil erosion at basin scale where human activities have altered the physical characteristics of the region. The Dongjiang River basin, an economically advanced and densely populated area in China, chiefly comprises six cities: Ganzhou, Heyuan, Huizhou, Dongguan, Guangzhou, and Shenzhen. The river is the primary water source for these cities as well as Hong Kong. In fact, the proportion of Dongjiang water in Hong Kong's annual water supply has steadily increased, from only 8.3% in 1960 to approximately 70% or even slightly greater than 80% in recent years (Jiang et al. 2007). It will have huge implications for economic development or even social stability if the water sources were to be affected by long-term pollution events (e.g., pollution caused by water and soil erosion). Moreover, the downstream region of the basin is the most developed area in Guangdong Province (even in China) after years of development. However, this region is currently facing severe ecological and environmental issues including soil erosion caused by tremendous changes in land cover (Liu et al. 2008; Wu & Chen 2012; Liu et al. 2015). A systematic study of soil erosion in the basin is urgently needed.

Therefore, the main objectives of this study are: (1) to propose a modeling approach of soil erosion using RUSLE and GIS tools and (2) to assess the soil erosion situation in the Dongjiang River basin from 1990 to 2010. The focal point of this study is to explore the spatial dynamic changes of soil erosion caused by land-cover changes at basin scale. The proposed approach can expediently evaluate the spatial heterogeneity and dynamic changes of soil erosion. This study also provides methodological tools for soil conservation planning and soil erosion prevention in the study area, which can be spread to similar regions.

Digital elevation model (DEM), daily precipitation data, soil type data, and land-cover data were used in this study. DEM data (SRTM30) were available from the CGIAR-CSI SRTM 30 m Database (http://srtm.csi.cgiar.org). Daily precipitation data of 34 rainfall observation stations (Figure 1) from 1960 to 2010 were accessed from the Hydrology Bureau of Guangdong Province (http://www.gdsw.gov.cn/wcm/gdsw/index.html) and the China meteorological data sharing service system (http://www.escience.gov.cn/metdata/page/index.html). Soil type data were obtained from the Food and Agriculture Organization of the United Nations (http://www.fao.org/home/en/). Land-cover data for 3 years (1990, 2000, and 2010) used for analysis in this study were provided by the Resources and Environment Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn/Default.aspx). The data-processing tools included ArcGIS 9.3 and MS Excel. The slope length and steepness factor were estimated from SRTM DEM (30 m). All the above-mentioned data were converted to the spatial resolution of 100 × 100 m, which means that the Dongjiang River basin consists of approximately three million grids.

In Equation (3), the value range of β is 1.2 to 1.8 and β = 1.5 is used in this study according to previous research in similar regions of China (Yu & Rosewell 1996a; Ma et al. 2014). In Equations (4) and (5), S_h (mm) is the precipitation amount of the next half year, and R_a (mm) is the average annual precipitation. The rainfall erosivity of each month is calculated using the daily precipitation from 1960 to 2010 for each meteorological station. This paper used the average annual erosivity value as the rainfall erosivity factor and it can be computed by accumulating the monthly rainfall erosivity. The rainfall erosivity factor data were placed into the attribute database for a specific meteorological station and then converted into grid format by the Kriging spatial interpolation method.

Both the factors of L and S were calculated using the raster calculator tool of Arc.GIS 9.3 according to Equations (7)–(10) and were merged as a LS factor.

The land cover and management practice factor C is used to reflect the effect of cropping and management practices on soil erosion rates in agricultural lands, and the effects of vegetation canopy and ground covers on reducing the soil erosion in forested regions (Renard et al. 1997). This factor considers the variability of the vegetation cover and methods of land management, reflecting their protective function to the topsoil (Xiao et al. 2015). The C factor is generally determined by the Normalized Difference Vegetation Index (NDVI) (Maria et al. 2009; Chen et al. 2011) or land-cover type (Xu et al. 2008; Pan & Wen 2014). NDVI tends to focus on the vegetation situation, but the land-cover type considers more comprehensive terrestrial situations, such as cultivated land, forest land, grassland, mudflat, urban land, construction land, bare land, and water body. Therefore, this paper used the land-cover type to estimate the C factor (Table 2) by referring to relevant studies in south China. This paper mainly referred to the studies of Chen et al. (2007) and Lu et al. (2011) as their study regions were close to the Dongjiang River basin and had similar natural conditions.

The support practice factor P (ranging from 0 to 1) is the soil-loss ratio with a specific support practice to the corresponding soil erosion with upslope and downslope tillage (Renard et al. 1997). The P factor is set to 1 in the regions without any soil and water conservation measures in general and to 0 in the regions where water and soil erosion will not occur. A lower P factor value indicates effective conservation practice at reducing soil erosion. In fact, an authoritative and universal method to determine the P factor has not been found so far, making this factor an uncertain factor in the RUSLE model. This paper referred to the experimental method of Lu et al. (2011) that considers both the land-cover type and the slope (Table 3). The study region of Lu et al. (2011) was close to the Dongjiang River basin and their results were based on the investigation of soil and water conservation, providing a reliable reference for this paper. Then, the P factor values were assigned to the raster for further analysis.

The land-cover variation related greatly to the increasing regional economy. A large area of paddy field, forest land, and grassland was invaded by urban land, construction land, and garden plots, especially in the downstream region. A typical example is that Shenzhen City used to be a small fishing village in south Guangdong Province three decades ago but now has become an overwhelming megacity, tremendously changing the original land cover. The expansion of urban land displays rapid urbanization and intensive human activities, meaning that humans have also changed the original ecosystem causing huge pressure on water and soil conservation.

There are 22 soil types in the Dongjiang River basin and the soil erodibility factor was calculated by Equation (6) based on the soil type data. The spatial distribution map of soil erodibility (Figure 3(b)) and soil types in Table 5 show that the main soil types include hydromorphic paddy soil, pitted lateritic red soil, page latosolic red soil, pitted yellow soil, pitted red soil, pitted yellow-red soil, yellow-page soil, and red-page soil, which in total account for approximately 94.10% of all the soil types. The K factor value ranges from 0.2216 to 0.416 and the average value is 0.2652 with a Cv of 0.1696. The soil erodibility factor values in the basin are overall relatively small, indicating that the topsoil in the Dongjiang River basin has some anti-erosion ability. An overall lower K factor was found in the north as compared to that in the southern region of the basin.

The slope length factor (L) and the steepness factor (S) were merged as a topography factor (LS) in this paper. The LS factor value ranges from 0 to 29.4, with an average value of 8.86. However, the Cv reached 0.8691, suggesting that spatial topography variation is considerable in this hilly basin. Figure 3(c) shows that the lower values are mainly distributed in the downstream regions and the higher values are mainly in the western part of the basin. An overall higher LS factor was found in the north than in the southern region.

As the potential annual soil erosion does not consider social factors, it only reflects the rainfall erosion vulnerability based on geographical and geological environment conditions but cannot reflect the actual situation of erosion in the Dongjiang River basin. A large value of potential soil erosion does not mean severe erosion will actually occur. For example, an appropriate irrigation system and soil management practices can greatly reduce soil erosion, demonstrating that these human actions are capable of reducing the potential effect to a certain extent. Human activities can be reflected by the land cover as displayed by the C and K factor in the RUSLE model. The land-cover types of 1990, 2000, and 2010 were selected and the influence of human activity was mainly focused on the dynamic changes of the land-cover type. As the six factors – rainfall erosivity factor R, soil erodibility factor K, slope length factor L, steepness factor S, cover and management practice factor C, and support practice factor K, have all been calculated, the soil erosion amount was calculated by Equation (1). The calculation results show that the soil erosion amounts ranges are 0 to 4,265.72, 0 to 4,265.72, and 0 to 4,999.69 t/(ha·year) in 1990, 2000, and 2010, respectively. The results were then classified into six levels according to the Standards for Classification and Gradation of Soil Erosion (SL 190–2007) published by the Ministry of Water Resources of the People's Republic of China in (2008). The classification standards, with a unit of t/(ha·year), were set as follows: Level I (tiny erosion), A < 5; Level II (mild erosion), 5 ≦ A < 25; Level III (moderate erosion), 25 ≦ A < 50; Level IV (intense erosion), 50 ≦ A < 80; Level V (more intense), 80 ≦ A < 150; and Level VI (extreme intense), A ≧ 150. The higher the level is, the greater soil erosion will be.

For different regions, the characteristic values of soil erosion are shown in Table 7. Evidently, the annual average value increases from upstream to downstream, suggesting that the soil erosion in the downstream region is more severe than that in the upstream region. In addition, the mean values increased from 1990 to 2010 among the three regions, including the whole basin. From upstream to downstream, the soil erosion situation becomes more serious from a space perspective. Meanwhile, erosion has slightly increased in the past two decades from a time perspective.

The above assessment and analysis indicate that the overall soil erosion situation in the Dongjiang River basin since 1990 was not severe as most of the areas belong to Level I and Level II. This is because, first, the Dongjiang River basin is located in the subtropical zones of south China. Abundant rain, warm temperature, and fertile land result in lush vegetation throughout the basin (Liu et al. 2010). Even if destroyed, the vegetation will quickly repair itself under the favorable environment. The self-healing feature of the vegetation is a necessity for conserving soil and water in the destructive and vulnerable areas. Second, numerous protection measures have been taken in the last two decades. For example, 45 first-class and 41 second-class protection zones were set up to protect the water sources (DRBA 2007). Unreasonable human activities in these zones, such as draining sewage, building factories, farming and grazing, have been under great restrictions, tremendously improving the resistance to water and soil erosion. In view of the serious soil erosion problems in some local areas, a large number of measures have been adopted by the Guangdong government, such as engineering measures, biological measures, and agrotechnical measures. The engineering measures mainly included building small reservoirs, planting shelter forests, and building terraced fields in the potential erosion areas. The biological measures, such as afforestation and conversion of farmland to forest or grass, were included in the ecologically vulnerable areas. Agrotechnical measures, such as sprinkler irrigation, plastic film mulching, rotation tillage, and inter-planting, were taken in the cultivation areas. These measures can effectively reduce water and soil loss. Therefore, both natural and human factors are applied to the mild erosion in the Dongjiang River basin on the whole.

Even so, the soil erosion situation requires more attention. First, the increased annual average values, areas of higher levels (Levels IV–VI), and areas that were converted from lower levels to higher levels indicate accelerated soil erosion from 1990 to 2010. Despite taking many measures, the rapid expansion of urban areas occupying a large amount of paddy fields, grasslands, and forestlands has greatly changed the characteristics of the original soil ecosystem conditions, causing an accelerated soil erosion process that adversely affects the conservation of water and soil. Second, the different intensity of human activities resulted in a great regional difference in the erosion degree across the basin. For example, the annual average value of the downstream regions reached 35.66 t/(ha·year) (Level III, moderate erosion) in 2010, but only 11.82 and 14.43 t/(ha·year) (Level II, mild erosion) in the up- and midstream regions. Some downstream areas, such as the western Huiyang, face a particularly serious erosion problem and some erosion sites even reached 4,999.69 t/(ha·year) (Level VI, extreme intense erosion). Finally, the pressure for water and soil conservation will be greater in the future due to the continuous and strengthening human activities. The cities in the mid- and downstream regions are presently eager to develop the economy, which may lead to more intensive land-cover changes. Similarly, the upstream regions will also face much more pressure by a developing economy in the future, posing great challenges to the basin's soil and water conservation. Therefore, the soil erosion situation requires attention at present and in the future. Given the important economic and political status in Guangdong Province, successful and efficient conservation measures are thus still highly imperative to safeguard water and soil resources in the Dongjiang River basin.

The study proposed an approach that links RUSLE and GIS tools to assess and map soil erosion at basin scale. The procedure was applied on the Dongjiang River basin (south China) where human activities have altered physical characteristics of the basin. The following conclusions are drawn from the study:

• The procedure linking RUSLE and GIS tools was found to be an efficient approach to assessing and mapping soil erosion on basin scale.

• Land-cover change in the Dongjiang River basin greatly impacted soil erosion and the erosion amount increased on the whole from 1990 to 2010 and the erosion in the downstream region is more serious than that in the mid- and upstream region.

• The total areas of Levels I and II (mild erosion) occupied approximately 92.9%, 92.4%, and 91.15% in 1990, 2000, and 2010, respectively, while the other high levels of erosion occupied less than 10% in the study region. The overall soil erosion situation in the Dongjiang River basin is mild but extreme intense erosion occurs in some local areas.

• Approximately 90.85% of the areas maintained the original levels of erosion, but 9.15% changed. Among the changed areas, 5.84% converted from lower levels to higher levels and 3.32% converted from higher levels to lower levels.

The gradual deterioration condition caused by land-cover changes at present or in the future requires more attention. These results are important in terms of soil conservation planning and prevention, the reduction of water and soil erosion, and other applications in the region, while the procedure is worth applying in other regions.

The research is financially supported by the National Natural Science Foundation of China (Grant No. 51210013, 51479216, 51579105), the water science and technology program of Guangzhou (Grant No. GZSWKJXM2012-02), the Public Welfare Project of Ministry of Water Resources (Grant No. 201201094, 201301002-02, 201301071), the project for Creative Research from Guangdong Water Resources Department (Grant No. 2011-11). The authors greatly appreciate the reviewers and the editors for their critical comments that greatly helped in improving the quality of this paper.