Introduction

Soil erosion is a complex geophysical process that occurs on terrestrial surfaces and involves the dispersion, detachment, transportation, and deposition of soil under the influence of external forces. This process serves to illustrate the complex interdependence and mutual influence between natural environments and human activities1. On an annual basis, soil erosion is responsible for the loss of between 25 and 40 billion tons of topsoil across the globe. This has a detrimental impact on land degradation and agricultural productivity, making soil erosion one of the most significant ecological challenges facing the planet2. The consequences of soil erosion extend beyond the accumulation of sediment in rivers, reservoirs, and lakes. They also encompass ecological concerns, including heightened sedimentation, deteriorating water quality, reduced biodiversity, and adverse effects on human well-being and safety3. The assessment of soil erosion represents a fundamental element in the accurate evaluation of the effectiveness of soil and water conservation initiatives, the assessment of their benefits, and the provision of essential insights for the formulation of strategies for soil conservation and national economic development4.

At present, research at both national and international levels has explored in great depth a range of aspects related to soil erosion and its consequences. This includes the examination of the underlying causes of soil erosion, the spatiotemporal patterns of erosion evolution, and the assessment of the sensitivity of different ecosystems to this process5. Among these methods, the quantitative analysis and evaluation of soil erosion represent pivotal areas of inquiry6,7. Scholar Liu Pan conducted a quantitative analysis of soil erosion intensity and the dynamic changes in its associated factors in Northeast China via a modified universal soil loss equation (USLE). The findings indicated that from 2010 to 2017 (assuming a future projection due to the likely typographical error in the year), the average soil erosion modulus in the study area decreased from 608.75 t/(km².a) to 678.25 t/(km².a), suggesting an improvement in soil erosion conditions. However, the latter value appears to be a typographical error or misunderstanding, as it is inconsistent with the observed trend of improvement, which would suggest a decrease in the erosion modulus8,9,10. Additionally, Gu Zhijia and colleagues evaluated soil erosion in the Manchuan and Mangang regions of Northeast China and documented a positive correlation between the slope gradient and the soil erosion modulus. Notably, they also observed that as slope length increased, the soil erosion modulus initially increased but subsequently decreased11,12,13.

In recent years, the rapid development of information technology has provided new opportunities for research in the field of soil erosion and land use. This research has gradually evolved towards diversification and multiscale directions14,15. Based on the RUSLE model, He Jiaying et al.16. A model of the response of soil erosion intensity to land use/cover change was constructed and it was reported that, as a consequence of the influence of different social factors during different periods, LUCC affects soil erosion and soil conservation. With the increase in the intensity of human irrational land use activities, the deterioration of soil properties has led to an increase in soil erosion, which in turn has restricted the structural mode of land use, resulting in a reduction in land productivity and a further intensification of the man–land contradiction17,18.

As a region with a predominantly agricultural economy, Lishu County has not been the subject of extensive research into soil erosion within its boundaries. This paper employs a quantitative analysis of soil erosion in Lishu County, leveraging data on precipitation, slope, soil, land use, and the normalized difference vegetation index (NDVI) from 2000 to 2020. The data have been organized into three decades and integrated with (GIS) to facilitate the RUSLE model’s application. The objective of this study is to elucidate the spatiotemporal dynamic characteristics of soil erosion in the county over the past two decades. The study explores the interrelationship between soil erosion and slope, employing the GCCM model to elucidate the coupling relationship between soil erosion and land use, and to quantify the contributions of diverse land use types to soil erosion. From the perspective of land management, the implementation of targeted prevention measures for soil erosion in Lishu County is proposed. This will serve to establish a robust foundation for the future of ecological conservation efforts in the region.

Study area overview and data sources

Overview of the study area

As illustrated in Fig. 1, Lishu County is located between 123°45’ and 124°53’ east longitude and 43°02’ and 43°46’ north latitude. The research region encompasses an area of 2925.5 square kilometres within Lishu County. The climate is temperate continental monsoon, characterised by both rainfall and high temperatures. The climate of the region is characterised by significant fluctuations in temperature across different seasons and time periods. In particular, the summer months, with July and August representing the peak period of precipitation, exhibit a distinct contrast in climatic conditions. The region’s topography slopes from west to east, with the Songliao Plain forming its western border. The terrain is predominantly flat, characterised by fertile black soil and chernozem, which has earned it global renown as one of the most productive “black soil zones” and a significant hub for commodity grain production19. Lishu County has a single land use type structure, with cultivated land accounting for 94.45% of the total land area, making it vulnerable to erosion20. In accordance with the directives of the Ministry of Agriculture and Rural Affairs and the Ministry of Finance, as outlined in the Action Plan for Conservation Tillage of Black Land in Northeast China (2020–2025), emphasis is placed on ecological precedence, balanced protection and development, and the holistic advancement of conservation tillage supported by scientific innovations. This approach aims to synchronise the preservation of black soil with sustainable agricultural development. Within this framework, Lishu County actively promotes the “Pear Tree Model“21. During a visit to Lishu County, General Secretary Xi Jinping emphasized the imperative to implement effective measures to safeguard the “giant panda of cultivated land”—black soil—and advocated for the comprehensive dissemination of the “Pear Tree Model” to wider extents22. In recent years, Lishu County has been honored with accolades such as the “National Agricultural Modernization Demonstration Zone,” “National Agricultural Green Development Pioneer Zone,” and “National Agricultural Science and Technology Modernization Pioneer County” conferred by various national ministries and commissions.

Fig. 1
figure 1

Overview of the study area. The map was created via ArcGIS V. 10.8 (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/resources) based on the standard map No. GS(2019)1822 from the Standard Map Service of the Ministry of Natural Resources (http://211.159.153.75/). No modifications were made to the base map.

Full size image

Data sources and Processing

The following section presents a comprehensive list of all the data employed in this study. The precipitation data were sourced from annual records collected from 53 meteorological stations across Jilin Province, with precipitation distribution maps generated through kriging interpolation. The land use data were categorised into six primary land types in accordance with the LUCC classification system, which was developed by the China Academy of Sciences. To guarantee the precision of the data, all raster data underwent a process of preprocessing. This involved the delineation of the research scope through the use of mask technology, the reprojection of the data onto the GCS-WGC-1984 coordinate system, and the uniform resampling of the data to a resolution of 300 m. This process provided a robust foundation for subsequent analyses. For details, refer to Table 1.

Table 1 Model data sources and utilization.
Full size table

Research methods

RUSLE model

The RUSLE model was developed by W. Wischmeier and D. Smith following a comprehensive analysis of observational data from over 10,000 runoff plots in 30 states across the eastern United States over the past three decades23. As our comprehension of the mechanisms underlying soil erosion has deepened and as we have sought to facilitate the pervasive application of computer technology in soil erosion prediction, the United States Department of Agriculture has implemented a revised model based on the Universal Soil Loss Equation (USLE) in 1997, with a broader scope of applicability. The RUSLE model has been widely employed in the quantitative calculation and analysis of soil erosion24. Nevertheless, there are several constraints associated with its implementation, including computational complexity and the challenge of acquiring model parameters. Notwithstanding the aforementioned limitations, geographic information systems (GISs) are equipped with robust spatial analysis capabilities, thereby providing substantial support in addressing the aforementioned challenges. With the advancement of “3S” (GIS, GPS, RS) technology, it provides data support for soil erosion research and enhances the efficiency of soil erosion assessment. The calculation formula is as follows:

$${\text{A}}={\text{R~}} \times {\text{K~}} \times {\text{LS~}} \times {\text{C~}} \times {\text{P}}$$
(1)

Here, A represents the annual soil erosion [t/km− 2。a], R denotes the rainfall erosivity factor [MJ mm/(hm² h a)], K signifies the soil erodibility factor [t hm² h/(MJ mm hm²)], LS stands for the slope length and gradient factor, C represents the cover and management factor, and P signifies the soil and water conservation factor. On the basis of precipitation data since 2000, soil data, and digital elevation model (DEM) remote sensing images, various factors influencing soil erosion were extracted through data calculation and analysis, and the soil erosion modulus and intensity grade of the study area were systematically analysed.

Soil erodibility factor

The soil erodibility factor (K) is a pivotal element in the RUSLE model, reflecting the susceptibility of the soil to erosion. In this study, the China soil dataset of the World Soil Database (HWSD) was employed for the purpose of determining the soil erodibility factor25. The data are presented in vector or raster form, and geographic information system (GIS) functions such as splicing, clipping, projection, attribute editing, and raster calculation are employed for data processing.Williams et al. (USDA, 1990) adopted the K factor estimation method from the EPIC (Erosion-Productivity Impact Calculator) model, which is based on the soil organic matter content and soil particle size composition and is expressed by the following formula:

$$\begin{aligned} ~ & K = 0.1317 \times \left\{ {0.2 + 0.3exp\left[ { - 0.0265SAN\left( {1 - \frac{{SIL}}{{100}}} \right)} \right]} \right\} \\ & \quad \times \left( {\frac{{SIL}}{{CLA + SIL}}} \right)^{{0.3}} \times \left\{ {1 - \frac{{0.25C}}{{C = exp\left( {3.72 - 2.95C} \right)}}} \right\} \\ & \quad \times \left\{ {1 - \frac{{0.7SN}}{{SN + exp\left( {22.9SN - 5.51} \right)}}} \right\} \\ \end{aligned}$$
(2)

Here, SAN represents the sand content (%), SIL denotes the silt content (%), CLA signifies the clay content (%), and C represents the organic carbon content in the soil (%).

Rainfall erosivity factor

Rainfall erosivity (R) represents the potential impact of rainfall conditions on soil erosion and serves as a crucial indicator for predicting soil erosion. In light of the study’s primary objective, which is to assess soil erosion in response to alterations in surface structures, such as changes in vegetation cover and land use patterns influenced by human activities, the rainfall data employed to calculate R are unified and restricted to the multiyear average rainfall from 2000 to 2020. In this study, precipitation data from 53 meteorological stations in Jilin Province over a 20-year period were used to calculate the R value via a precipitation distribution map generated by kriging interpolation, with the resolution resampled to 300 m to maintain consistency with other data. The calculation of the R factor is conducted in accordance with the guidelines set forth by the Ministry of Water Resources of the People’s Republic of China26. The formula for this calculation is as follows:

$${R_n}=0.053 \times P_{n}^{{1.665}}$$
(3)

Here, \(\:{\text{R}}_{\text{n}}\) represents the annual rainfall erosivity factor, and Pn denotes the annual rainfall.

Vegetation coverage factor

The vegetation coverage factor (C) serves to mitigate soil erosion. A value of 1 indicates complete bare ground with no vegetation, whereas a value approaching 0 indicates optimal vegetation coverage. The initial step involves the acquisition of vegetation coverage (FVC) data from remote sensing images. This is followed by the calculation of the vegetation coverage factor, which is based on the aforementioned vegetation coverage data. The processing of data is achieved through the utilisation of GIS functions, including the application of techniques such as splicing, clipping, projection, raster calculation, and conditional analysis. Building upon the foundations of previous research, this study employs algorithms developed by Cai Chongfa and colleagues to calculate the corresponding relationship between the FVC and the C value27, as expressed by the following formulas:

$$FVC=\frac{{NDVI - NDV{I_{min}}}}{{NDV{I_{max - NDV{I_{min}}}}}}$$
(4)
$$C=\left\{ {\begin{array}{*{20}{c}} 1&{c=1} \\ {0.6508 - 0.3436lnfvc}&{0<c \leqslant 0.783} \\ 0&{c<0.783} \end{array}} \right.$$
(5)

Here, FVC represents the vegetation coverage, and C denotes the vegetation coverage factor.

Soil and water conservation measures

The soil and water conservation measures factor represents the ratio of erosion on sloped land with soil and water conservation measures to erosion without such measures under equivalent conditions. This reflects the extent to which soil and water conservation measures are effective in mitigating soil erosion28.The factor varies between 0 and 1, with lower values indicating a lesser degree of erosion and a value of 1 denoting the absence of soil and water conservation measures. This factor is determined on the basis of land use data, which is typically presented in vector or raster formats. In order to utilise the data, it is necessary to employ GIS functions such as splicing, clipping, projection, attribute editing, and conditional analysis. In particular, the factor assigned to woodland is 0.90, while water areas are assigned a factor of 0. Cultivated land is assigned a factor of 0.30, grassland is assigned a factor of 0.50, bare land is assigned a factor of 1.00, and construction land is assigned a factor of 0.20. For further details, please see Table 2.

Table 2 Land use type assignment table.
Full size table

Slope length and slope factor

The slope length and gradient factor (LS) serve to illustrate the impact of topographic fluctuations on soil erosion29,30, which can be calculated from geomorphological parameters. The digital elevation model (DEM) data employed in this study were sourced from the Geospatial Data Cloud website and have a spatial resolution of 90 m. The slope length was calculated using a series of operations, including the filling of depressions and the calculation of flow directions, in accordance with the following formula:

$$L={\left( {\frac{\lambda }{{22.13}}} \right)^m}$$
(6)
$$\lambda =l \times cos\alpha$$
(7)

where \(\:\text{L}\) is the slope length factor; \(\:{\uplambda\:}\) is the horizontal projection slope; \(\:\text{L}\) is the slope of the surface water flow along the flow direction; \(\:{\upalpha\:}\:\) is the slope value of the water flow area; \(\:\text{m}\) is a variable slope index, and when \(\:{\uptheta\:}\)<0.57°, \(\:\text{m}\)=0.2; when 0.57°≤\(\:{\uptheta\:}\)<1.72°, \(\:\text{m}\)=0.3; when 1.72°≤\(\:{\uptheta\:}\)<2.86°, \(\:\text{m}\)=0.4; and when 2.56°≤\(\:{\uptheta\:}\), \(\:\text{m}\)=0.5.

The slope calculation is based on grading, using the McCooLDK formula below 10 degrees and the Liu constant rate formula above 10 degrees, with the following formula:

$$S=\left\{ {\begin{array}{*{20}{c}} {10.80 \times sin\theta +0.03}&{\theta <5^\circ } \\ {16.80~ \times sin\theta - 0.50}&{5^\circ <\theta \leqslant 10^\circ } \\ {21.91 \times sin\theta - 0.96}&{\theta >10^\circ } \end{array}} \right.$$
(8)

where \(\:\text{S}\) is the slope factor and where \(\:{\uptheta\:}\) is the slope.

2.1.6 Soil erosion modulus calculation.

The raster maps of the aforementioned factors have been set with a unified projection and grid resolution. The soil erosion modulus has been obtained through the superimposition operation of formula (1). The intensity of soil erosion is classified according to the soil erosion standard in the black soil area into six levels: micro, mild, moderate, intense, extremely intense, and severe erosion. The classification results of the soil erosion intensity in the study area for the years 2000, 2010, and 2020 have been obtained and are presented in Table 3.

Table 3 Classification Standard of Soil Erosion Grade [t/km− 2。A].
Full size table

GCCM model

GCCM is a model designed for the purpose of spatial causal reasoning, which identifies causal relationships between spatial cross-sectional variables. This is achieved by reconstructing the state space and employing cross-mapping prediction. The method is capable of reliably detecting the direction and strength of causal connections in systems exhibiting weak to moderate coupling. Furthermore, it offers significant advantages in the handling of inseparability and nonlinear relationships, thereby markedly enhancing causal inference within complex Earth systems. This model employs a combination of multiple observations to construct the state space of a dynamic system, thereby enabling the determination of whether two variables originate from the same dynamic system and, consequently, the establishment of their causal relationship31. The model formula is as follows:

$${\hat {Y}_s}\left| {{M_x}} \right.=\mathop \sum \limits_{{i=1}}^{{L+1}} \left( {{w_{si}}{Y_{si}}{\text{|}}{M_x}} \right)$$
(9)

where S represents the spatial unit for which the value of \(\:\text{Y}\) needs to be predicted and where \(\:{\widehat{\text{Y}}}_{\text{s}}\) is the predicted result. L is the embedded dimensional data used for prediction, and \(\:{\text{Y}}_{\text{s}\text{i}}\) is the predicted value at \(\:{\text{S}}_{\text{i}}\), which is also the first component of a state in\(\:{\:\text{M}}_{\text{x}}\), denoted as ψ(y,\(\:{\text{S}}_{\text{i}}\)). Furthermore, ψ(y, si) is determined by its one-to-one mapping point ψ(x,\(\:{\text{s}}_{\text{i}}\)), and ψ(x, si) is one of the L + 1 nearest neighbors of the focal state ψ(x, s) in \(\:{\text{M}}_{\text{x}}\).

Manifold prediction31 involves finding the E + 1 nearest neighbor points \(\:{\text{S}}_{\text{x},\text{i}}\) from \(\:{\text{S}}_{\text{x}}\) near to far in the \(\:{\text{S}}_{\text{x},{\text{i}}_{1}}\) and \(\:{\text{S}}_{\text{x},{\text{i}}_{2}}{，.\text{S}}_{\text{x},{\text{i}}_{\text{E}+1}}\) manifold and then calculating the Euclidean distance between \(\:{\text{S}}_{\text{x},{\text{i}}_{1}}\), \(\:{\text{S}}_{\text{x},{\text{i}}_{2}}{，.\text{S}}_{\text{x},{\text{i}}_{\text{E}+1}}\) and \(\:{\text{S}}_{\text{x},\text{i}}\) to obtain the weight:

$$weight\left( {\psi \left( {x,{s_i}} \right),\psi \left( {x,s} \right)} \right)={\text{exp}}\left( { - \frac{{dis\left( {\psi \left( {x,{s_i}} \right),\psi \left( {x,s} \right)} \right)}}{{dis\left( {\psi \left( {x,{s_1}} \right),\psi \left( {x,s} \right)} \right)}}} \right)$$
(10)

where exp represents the exponential function and where \(\:\text{d}\text{i}\text{s}\)() denotes the distance function between two states in the shadow manifold.

The calculation of the causal strength involves the following formula:

$$\rho =\frac{{{\text{Cov}}\left( {{\text{Y}},\hat {Y}} \right)}}{{\sqrt {{\text{Var}}\left( {\text{Y}} \right){\text{Var}}\left( {\hat {Y}} \right)} }}$$
(11)

where \(\:{\uprho\:}\) is the strength of causality, \(\:\text{C}\text{o}\text{v}\)() represents covariance, and \(\:\text{V}\text{a}\text{r}\)() represents variance. For raster data, the window size is used to represent the size of the library. Sugihara et al. proposed using the convergence of \(\:{\uprho\:}\) to infer causality. In the case of GCCM, convergence implies that \(\:{\uprho\:}\) increases with the library size and is statistically significant when the library is at its largest32.

The self-weight distribution33 represents the internal contribution relationship of each indicator. Through self-weight analysis and calculation, the self-weight distribution relationship of indicators in different regions can be clearly expressed. The calculation method is as follows:

$${w_i}={z_i}~/~\Sigma {z_i}$$
(12)

where \(\:{\text{z}}_{\text{i}}\) is the normalized value of each factor, and the sum of the normalized values of each factor is denoted as \(\:{\Sigma\:}{\text{z}}_{\text{i}}\).

Results and analysis

Spatial and temporal distribution characteristics of soil erosion in Lishu County

The distribution of soil erosion in Lishu County from 2000 to 2020 exhibited a yearly varying trend with a high degree of precision, as evidenced by rigorous academic and professional analysis. The predominant characteristics of soil erosion were mild and moderate levels, with instances of stronger erosion observed in only a few areas. Over the twenty-year period, the area experiencing mild erosion exhibited a gradual decrease in the central region of Lishu County. Conversely, the area affected by moderate erosion demonstrated a steady increase, with its expansion displaying a relatively concentrated trend, radiating outwards and indicating notable erosive capabilities and diffusion. The areas subject to moderate and higher-grade soil erosion initially exhibited an increase, but subsequently demonstrated a decline over time. Specifically, prior to 2010, these erosion-prone areas demonstrated an upward trajectory, whereas after 2010, there was a consistent annual decline. Furthermore, there was a discernible shift in the geographical distribution of erosion, with a gradual transition from the northern regions to the central areas of the study zone, manifesting as a punctate distribution (Fig. 2).

Fig. 2
figure 2

Spatial distribution of soil erosion in Lishu County from 2000–2020.

Full size image

Table 4 below demonstrates that between the years 2000 and 2020, the proportions of micro- and light erosion areas in Lishu County exhibited a declining trend, whereas those of moderate and severe erosion areas increased. However, the proportions of extremely severe and severe erosion were insufficient to warrant focused analysis. The rate of change for microeroded areas was − 4%, with its proportion reaching a maximum of 6% in 2000 before declining to 1% in 2010. In contrast, the alteration in light erosion was more pronounced, reaching − 28%. This was accompanied by a significant decline in its prevalence, from 89% in 2000 to 51% in 2010. This transition highlights the alarming deterioration in soil erosion conditions in Lishu County in recent years. Furthermore, the proportion of moderate erosion reached its lowest point in the year 2000, at 4%, and then increased steadily until it reached 36% by the year 2010. Similarly, the proportion of severe erosion reached its peak in 2010 at 8% before declining to 5% in 2020. Despite fluctuations in the proportions of moderate and severe erosion, both exhibited an overarching pattern of initial increase followed by subsequent decline, with change rates of 27% and 5%, respectively. Overall, the soil erosion situation in Lishu County has been exacerbated, with slight and moderate erosion being the predominant erosion type.

Table 4 Proportion and change rate of the soil erosion area in Lishu County from 2000–2020.
Full size table

As illustrated in Fig. 3, soil erosion in Lishu County has undergone a significant degree of transition. A total area of 11.35 km² within the research area exhibited a transition in the level of erosion, shifting from a high grade to a low grade. The most prevalent transitions were those of a mild and moderate nature. Moderate erosion was the dominant type from 2000 to 2010, with the majority of cases of moderate erosion originating from mild erosion. This trend indicates an increase in the degree of soil erosion and a deterioration in the regional erosion scenario. Furthermore, Lishu County has undergone a transition from low- to high-grade erosion, with 1211.89 km² of micro- and light erosion shifting to higher erosion grades. It is noteworthy that between the years 2010 and 2020, the area undergoing this transition amounted to 537.88 km², which indicates a considerable degree of advancement in the implementation of soil erosion control measures in Lishu County.

Fig. 3
figure 3

Soil erosion transfer matrix in Lishu County from 2000–2020.

Full size image

Relationships between soil erosion and land use types in Lishu County

As illustrated in Fig. 4, land use is a reflection of humanity’s capacity to adapt to and alter the natural environment. It demonstrates how humans adjust and shape their surroundings on the basis of natural and socioeconomic conditions in order to meet various needs and objectives34. The impact of precipitation and surface runoff on erosion is influenced by modifications to surface soil and alterations to ground vegetation coverage, as a result of changes in land use35. The assignment of a value of 0 to water areas during the calculation of the P factor has the effect of significantly limiting the extent of unused land in this region. The objective of this study is to analyse the correlations between various land use types, including cultivated land, forestland, grassland, construction land and soil erosion. The analysis demonstrated that the principal land use types linked with soil erosion in Lishu County are cultivated land, construction land, grassland, and forestland, in that order. A total of 93.4% of the soil erosion area in Lishu County is attributed to cultivated land, while construction land accounts for 6.2%. Soil erosion on cultivated land is predominantly mild, accounting for 59.76% of the total cultivated land erosion area. Following mild erosion, moderate and severe erosion cover 859.43 km² and 146.44 km², respectively. Similarly, soil erosion on construction land is characterised mainly by slight erosion, with moderate erosion being less prevalent. While slight erosion persists in grasslands, the extent of this phenomenon is relatively limited, accounting for a mere 0.25% of the total soil erosion area. The forestland area exhibits a similarly low erosion rate, with an area of only 4.56 km². It is noteworthy that, as illustrated in Fig. 5, both extreme intensity and intense erosion are absent in grassland and woodland soils. It seems probable that this phenomenon can be attributed to the high level of vegetation coverage and the favourable ecological conditions that prevail in these areas. This is facilitated by effective government management practices that are aimed at enhancing soil erosion resistance.

Fig. 4
figure 4

Soil erosion area under different land use types.

Full size image
Fig. 5
figure 5

Soil erosion areas of different grades under different land use types.

Full size image

Relationships between soil erosion and slope in Lishu County

The slope of a given area is a critical factor in the process of soil erosion, and is inextricably linked to the intensity of this process. This study employs the classification standard for slope erosion in black soil areas in order to accurately categorise slopes into six grades36,37, thus facilitating an analysis of erosion changes under different slope conditions. In the 0° to 3° range, the erosion area is 484.83 km², with light erosion covering 409.03 km², representing the highest proportion at 84.36%. The extent of erosion observed in this range is relatively limited, with only moderate and above-moderate intensity erosion occurring. In the 3° to 5° range, the erosion area spans 1308.35 km², with micro, mild, moderate, and intensive erosion accounting for 39.97%, 33.63%, 16.75%, and 8.73%, respectively. The combined effect of extreme and intensive erosion was found to account for less than 1% of the total erosion. In the 5° to 8° range, the erosion area is 836.05 km², with erosion proportions similar to those observed in the 3° to 5° slope region (37.35%, 36.9%, 15.49%, and 1%, respectively). In the 8° to 15° range, the erosion area is 239.12 km², with slight erosion accounting for 29.74% and 52.73%, respectively. Moderate and above-moderate intensity erosion account for 17.53% of the total erosion area. The erosion area within the 15° to 25° range constituted 1.78% of the total erosion area. Erosion in areas greater than 25° is less than 1%. Further details can be found in Fig. 6.

Fig. 6
figure 6

Soil erosion area of different grades under various slope categories.

Full size image

From 2000 to 2020, microerosion occurred primarily in areas of less than 8°, covering 971.35 km², which constituted 33.2% of the total erosion area. Areas that exhibited only slight erosion were 1304.97 km², representing 44.61% of the total area, and were primarily distributed in areas of less than 15°. The area of moderate erosion amounted to 410.81 km², representing 14.04% of the total erosion area. The majority of this moderate erosion occurred within the 3° to 5° range. The area of intensive erosion is 213.74 km², distributed widely across the 3°–8° range, accounting for 7.3% of the total erosion area. The collective area of extreme intensity and intense erosion is less than 1% of the total area. Further details can be found in Table 5.

Table 5 Areas of soil erosion grades across different slopes.
Full size table

Coupling relationships between soil erosion and land use type

The GCCM elucidates the causal relationship between soil erosion and land use types. Figure 7 depicts the soil erosion and land use data following processing, exhibiting consistent resolution, projection, and grid row and column numbers. Figure 8 illustrates the cross-prediction output between soil erosion and land use types. Once the maximum inventory capacity (L) has been reached, the correlation coefficient (ρ) of the mapping function y xmap x is found to be 0.38. The results of the experiment indicate that the coupling results consistently exceed 0.15. A higher numerical value (y xmap x) is indicative of a more significant causal effect of land use type on soil erosion. Conversely, for the mapping function x xmap y, the maximum value of ρ is 0.36, whereas the minimum value remains at 0.095. The value of this mapping function represents the inverse causal effect of soil erosion on land use. In light of the comprehensive experimental evidence, we put forth the hypothesis that there are robust positive and negative causal relationships between soil erosion and land use.

Fig. 7
figure 7

Soil erosion and land use data processing diagram.

Full size image
Fig. 8
figure 8

Coupling Results between Soil Erosion and Land Use Types.

Full size image

By means of calculation, the self-weight distribution of each index factor is determined. The cultivated land index has a self-weight ratio of 78.57%, with cultivated land occupying as much as 94.45% of Lishu County’s area. The single structure of land use types and the extensive presence of agricultural land imply that a large area of soil is exposed without the protection of natural vegetation37. Moreover, long-term agricultural practices have been shown to alter the physical and chemical properties of the soil. The use of chemical fertilisers and pesticides has also been identified as a potential source of soil contamination, which can further compromise soil resilience. This has a direct impact on agricultural production and economic development in Lishu County. The self-weight ratio of the construction land index is 18.42%. For further details, please refer to Table 1. The processes of urbanisation and rural revitalisation give rise to disturbances and excavations of the original surface, which in turn disrupt the surface vegetation and the natural structure of the soil. Concurrently, the advancement of water conservancy construction has mitigated soil erosion to a degree by regulating surface runoff. The self-weights of the remaining indicators are relatively insignificant and exert a minimal influence on the process of soil erosion. For further clarification, please refer to Table 6.

Table 6 Areas of soil erosion grades with different slopes.
Full size table

Discussion and conclusion

Discussion

Analysis of spatial and temporal dynamic changes of soil erosion

Between the years 2000 and 2020, the soil erosion status in Lishu County exhibited a 28% decrease in mild erosion, a 28% increase in moderate erosion, and a 6% increase in erosion above intense levels. The latter were primarily concentrated in the central region. The spatial distribution of soil erosion in Lishu County is characterised by an interlaced pattern of patches and points (Fig. 3), with the highest erosion intensity observed in the northern and central regions compared to other areas. The low-lying topography of Lishu County, coupled with minimal elevation changes, renders the region susceptible to erosion, with variations in vegetation cover and land use types exerting a predominant influence on the observed changes. As a prominent agricultural county, Lishu County boasts a cultivated land area that constitutes 94.45% of its total area, with construction land accounting for 4.5%. This distribution is characterised by a spotted pattern, as illustrated in Fig. 2. The results of the RUSLE model indicate that the average soil erosion rate in Lishu County is 2082.28 t/km²/a, which is in close alignment with the findings of other scholars in this field. As illustrated in Table 4, the area experiencing moderate erosion demonstrated a persistent increase, reaching 36% prior to 2010. However, subsequent to this period, a declining trend was observed. Moreover, although the intense erosion area peaked before 2010 and subsequently declined, this trend concurs with the research by Liu Pan et al. on soil erosion grade changes in Northeast China from 2010 to 2017, which revealed a “two increases and four decreases” pattern, namely, increases in micro- and mild erosion and decreases in the area and proportion of moderate and above-moderate erosion8.

Correlation analysis between soil erosion and land use types

The dynamic evolution of soil erosion in Lishu County is closely associated with alterations in land use types. The predominant land use type in Lishu County is cultivated land, with construction land represented in a dispersed, spotty pattern within the study area (Fig. 7). According to Fig. 6, the comparison of soil erosion intensity among different land use types in Lishu County revealed the following order: cultivated land > construction land > grassland > forestland. From 2000 to 2010, the degree of soil erosion in cultivated land tended to intensify, shifting from mild erosion to moderate erosion. However, following the year 2010, a notable improvement in soil erosion was observed, with a 6% reduction in areas experiencing moderate and above-moderate erosion. In conclusion, the excessive cultivation of land in Lishu County is likely to have resulted in a reduction in soil organic matter content. Furthermore, the application of chemical fertilisers and pesticides has been identified as a contributing factor to soil compaction and hardening, which in turn has accelerated soil erosion during this period. In response to the soil erosion situation in Lishu County, research institutions spearheaded the initiative to implement conservation tillage in 2008, establishing an experimental base to that end38. By establishing an expert guidance team, deploying long-term monitoring sites and strengthening basic research, a highly efficient modern agricultural scientific research and extension system was successfully created. This system provides scientific support for the country to take the lead in agricultural modernisation. By 2020, the area protected by black soil in Lishu County had reached 20 million acres39,40.

Coupling relationship analysis

The coupling results of the GCCM model between the soil erosion modulus and land use type data indicate a clear causal and reverse causal relationship between soil erosion and land use types in Lishu County. A self-weighting analysis of the contributions of various land use type indicators to soil erosion in Lishu County revealed that cultivated land and construction land play significant roles in the dynamic changes in soil erosion in the county. On the basis of this premise, we must not only focus on cultivated land and construction land as entry points to curb soil erosion through scientific and rational land use planning but also actively adjust land use layouts from the perspective of soil erosion control, thereby promoting economic development in Lishu County.

Explanation of model adaptability

This study employs the Revised Universal Soil Loss Equation (RUSLE) model for soil erosion modelling. Each factor in the model can be more specifically represented in accordance with the conditions prevailing in different regions. Once the slope length and steepness factor have been calculated, a segmented formula is employed, thereby facilitating a more comprehensive examination of the surface characteristics of Lishu County. Furthermore, land use type exerts a predominant influence on soil erosion, while land use patterns serve as the primary drivers of spatiotemporal dynamic changes in soil erosion41. The selection of the R-factor calculation method is based on a comparison of the results obtained from the classic precipitation algorithm and the daily precipitation fitting algorithm, as conducted by Si Zhiwen and colleagues in Jiangxi Province. The calculated results of rainfall erosivity between the two algorithms exhibit a high degree of similarity, with minimal average discrepancies42. In consideration of the fact that the study area in Lishu County is characterised by a small monsoon climate with high and concentrated precipitation, a uniform annual average rainfall from 20 time series spanning the period from 2000 to 2020 is employed for the purpose of analysing the precipitation factor.

Conclusion

1. From 2000 to 2020, soil erosion in Lishu County was predominantly characterised by mild and moderate erosion, with areas of very slight and slight erosion exhibiting a 4% and 28% reduction, respectively. The incidence of moderate, intense, and extremely intense erosion increased by 27%, 5%, and 1%, respectively. During this period, the trend of soil erosion exhibited a fluctuating pattern, characterised by periods of increasing erosion, decreasing erosion and minimal change. The areas exhibiting reduced erosion were predominantly distributed in small patches within the southern portion of the study area, whereas the areas displaying increased erosion were concentrated in the central region.

2. The land use in Lishu County is dominated by cultivated land and forestland, with the soil erosion intensity following the order of cultivated land > construction land > grassland > forestland. From 2000 to 2020, the area of soil erosion in regions where cultivated land was present accounted for 93.4% of the total area of soil erosion in Lishu County. In contrast, the area of soil erosion in regions where construction land was present accounted for only 6.2% of the total area of soil erosion.

  1. 3.

    The occurrence of soil erosion in Lishu County displayed a discernible pattern that was associated with changes in slope gradient. The erosion area in the study area is observed to decrease in a gradient-dependent manner towards both sides, with a central concentration at approximately 3°–5°, which is characterised by a prevalence of very slight and slight erosion. In light of the extensive cultivation of land in Lishu County and the prevalent erosion issues affecting sloping farmland, the “Lishu Model” has been instrumental in the implementation of effective control measures.

  2. 4.

    The results of the relevant analyses indicate that soil erosion in Lishu County is closely related to local land use patterns and surface slope characteristics. Consequently, there is a need to intensify future soil and water conservation efforts in the county. It is essential to ensure the continued maintenance of existing soil and water conservation achievements, while concurrently implementing effective management strategies in key areas. In order to effectively address soil erosion in Lishu County, it is essential to identify the underlying causes and implement a comprehensive approach that integrates ecological restoration and artificial management. Given its status as a principal national commodity grain base, the county should accord the highest priority to the restoration of farmland to its original state as forests and grasslands, thereby capitalising on the intrinsic self-regeneration capabilities of the natural environment. In areas exhibiting severe erosion, enclosed or semi-enclosed management strategies should be employed. Moreover, public awareness of soil and water conservation provides a robust foundation for conservation initiatives.