Accurately extracting watershed boundaries is critical for hydrological modeling and environmental management. Traditional extraction methods from Digital Elevation Models (DEMs) rely on manually defined thresholds and supplementary terrain features, limiting adaptability and efficiency. To address these issues, this study developed a watershed boundaries extraction framework based on a Residual Bottleneck Attention Multi-feature Fusion Network (RBM-SegNet). The framework consists of three components: an input layer, a semantic segmentation model, and a post-processing module. Key contributions include: (1) utilizing the [DEM, Slope, Hillshade, and Aspect] functions as the optimal input combination; (2) introducing residual connections and the Bottleneck Attention Module (BAM) to enhance feature transmission and suppress irrelevant regions; (3) incorporating multi-feature fusion to refine structural and detail prediction; and (4) incorporating post-processing to improve output-completeness and hydrological consistency. The experimental results show that RBM-SegNet outperforms traditional and existing deep learning methods in accuracy, demonstrating strong potential for practical applications.

Slope length and steepness (LS) factors derived from Digital Elevation Models (DEMs) are critical for soil erosion modeling. However, acquiring large-scale, high-resolution DEMs remains challenging. Existing downscaling methods often fail to integrate global and local terrain features or recover high-frequency details in regions with large slope variations. This study proposes a terrain feature-aware downscaling model based on MambaIR (TfaDM_MambaIR), designed with large kernel attention module (LKA_Block) to capture global terrain structures and a texture detail feature extraction module (Res_MambaIR) to model long-range spatial dependencies in elevations and enhance fine-grained terrain relief, effectively reconstructing continuous terrain structures and complex details. A collaborative loss integrates explicit constraints on slope steepness and terrain structure lines to optimize global elevation precision while preserving complex topographic features. Experiments show that TfaDM_MambaIR reduces DEM/LS errors by 21%-80%/1%-41% and improves DEM/LS PSNR by 3%-25%/4%-15%, providing a feasible model for improving DEM and LS reliability at large scales.

With the intensification of climate change and the increased frequency of extreme weather events, terraces, as a traditional agricultural practice, are facing severe challenges. This study investigated the characteristics of soil erosion on terraces under extreme rainstorm conditions, the dominant controlling factors influencing soil erosion, and the relationships between these factors in the terrace area of Qiugou watershed on the Loess Plateau. Utilizing high-precision digital elevation model (DEM) data acquired through field surveys and unmanned aerial vehicle (UAV) aerial photogrammetry, topographic and hydrological factors were extracted through visual interpretation and model simulations. High-resolution analysis was carried out on terrace erosion types, soil erosion intensity and the relationships between topographic/hydrological factors and runoff generation/erosion processes during extreme rainstorms. The results indicate that the extreme rainstorm caused five primary forms of terrace damage in the Qiugou watershed: gully erosion, cave erosion, rill erosion, slumping/collapse and shallow landslides. Within the study area, the soil erosion intensity of terrace platform and terrace wall, including channelized erosion and gravitational erosion, were 3732.10 and 27 843.26 t/km2, respectively. The sheet erosion intensities of terrace platform and terrace wall were 2898.56 and 20 143.02 t/km2, respectively. Terrace wall is the primary source of soil erosion in terraces. New terraces suffered significantly more damage, with a soil erosion intensity 1.16 times higher than that of old terraces. Slope gradient emerged as the key factor controlling the terrace soil erosion intensity, which increased with steeper slopes (reaching a peak in the 30 degrees-50 degrees range). Management strategies should therefore prioritize areas with slopes between 30 degrees and 50 degrees, which are commonly distributed within terrace wall. Optimizing terrace design to reduce soil and water loss requires synergistic regulation by combining the LS factor (slope length and steepness factor) and specific catchment area.

Gully erosion severely threatens land resources and agricultural sustainability, yet the role of subsurface erosionresistant soil layers remains poorly understood. This study integrated sub-meter imagery with stratified soil sampling (0-120 cm depth) across 79 gullies with comparable slopes gradient and catchment areas on a farm (570 km2) in the black soil region of northeast China to quantify how erosion-resistant layers control gully sidewall expansion and headcut retreat. Ten soil properties were analyzed to construct a PCA-based comprehensive soil erosion resistance score (CRS), revealing a decline in CRS with profile depth and a hierarchy of soil resistance: Black soil > Black soil-Loess transition layer > Loess > Loess-Sand transition layer > Fluvial sandy. The first two layers were identified as erosion-resistant layers. Through threshold effect analysis, a threshold erosion-resistant layer thickness of 53.88 cm was identified (p < 0.01) for sidewall expansion, revealing a 1.36 cm/yr acceleration in gully sidewall expansion per 1 cm thinning within the threshold thickness. Gullies with erosion-resistant layers below the sensitivity thickness experienced 2.03 times higher expansion rate. Gullies newly formed since 2010 exhibited a lower threshold (34.90 cm) than the pre-existing gullies. Gully headcut retreat rate was 43 % higher if the resistant layer was thinner than 54.28 cm, despite no significant detectable threshold. The threshold erosion-resistant layer thickness is potentially modulated by the depth of soil cracks and needs further investigation. This study highlights the importance of soil-profile features, not just surface properties, in gully erosion research. Integration of this threshold into gully erosion models could revolutionize gully prediction and precision conservation strategies.

The degradation of ‘black soil beach’ (BSB) ecosystems in the Three-River-Source region, characterized by widespread bald patches and severe soil erosion, poses a critical threat to regional ecological security and sustainable pastoralism. This study aims to elucidate the spatial distribution patterns and driving factors of bald patches in BSB degraded grasslands within the Guoluo Tibetan Autonomous Prefecture, providing a scientific basis for targeted restoration strategies. Utilizing multi-source remote sensing data (Landsat 8–9 OLI, UAV imagery, and Google Earth), we employed the Multiple Endmember Spectral Mixture Analysis (MESMA) method to identify bald patches, combined with the landscape pattern index and spatial autocorrelation to quantify their spatial heterogeneity. Geographical detector analysis was applied to assess the influence of natural and anthropogenic factors. The results indicate the following: (1) The patches are bounded by the Yellow River, showing a distribution pattern of ‘high in the west and low in the east’. The total area of patches reached 32,222.11 km2, accounting for 43.43% of the total area of Guoluo Prefecture, among which Maduo County and Dari County had the highest degradation rate. (2) With the aggravation of degradation, the patch density of each county increased first and then decreased, while the aggregation index and landscape shape index continued to decrease. (3) Spatial autocorrelation of bare patches strengthens with degradation severity (Moran’s I index 0.6543→0.7999). LISA identified two clusters: the high–high agglomeration area in the north of Maduo–Dari and the low–low agglomeration area in the southeast of Jiuzhi–Banma, revealing the spatial heterogeneity of the degradation process. (4) The spatial distribution pattern of bare patches was mainly affected by the annual average precipitation and actual stocking capacity, and the synergistic effect was significantly higher than that of a single factor. The combination of a 4491–4708 m high altitude area, 0–5° gentle slope zone, and soil texture (clay 27–31%, silt 43–100%) has the highest degradation risk. This multi-factor coupling effect explains the limitations of traditional single factor analysis and provides a new perspective for accurate repair.

Gully erosion susceptibility (GES) mapping is crucial for controlling gully erosion hazards and has become a significant focus of global research and management efforts. Machine learning models have proven effective in this field. However, in areas with different terrain complexity, the model shows significant variation in optimal resolution and algorithms, factor importance and spatial distribution of the model results, which limits their broader application. This study compares GES mapping in two small watersheds: one located in the complex terrain of the Loess Plateau and the other in the relatively flat terrain of the Northeast China Mollisol region. The model predictive accuracy was evaluated using 30% of the datasets that were excluded from model training. The results revealed that: 1) significant differences in optimal resolution of GES mapping in the two regions, which were 1-2.5 m for the Mollisol region, and 2.5-5 m for the Loess Plateau. The extreme boosting tree (XGBoost) algorithm achieved the best simulation results compared to random forest (RF) and gradient boosting decision tree (GBDT) in both regions. 2) Slope gradient and contributing area influenced gully distribution in both watersheds, with land use being critical in the Loess Plateau and distance from streams more important in the Mollisol region. 3) In the Loess Plateau watershed, 25% of the area was highly susceptible to gully erosion, while only 1% of the Mollisol watershed was highly susceptible. This research compared GES mapping in two watersheds with different terrain complexity, which would be beneficial for better use of machine learning in gully research.

The Loess Plateau is one of the most severely affected regions by soil erosion in the world, with a fragile ecological environment. Vegetation plays a key role in the region’s ecological restoration and protection. This study employs the Geographical Detector (Geodetector) model to quantitatively assess the impact of natural and human factors, such as temperature, precipitation, soil type, and land use, on vegetation growth. It aims to reveal the characteristics and driving mechanisms of vegetation cover changes on the Loess Plateau over the past 26 years. The results indicate that from 1995 to 2020, the vegetation coverage on the Loess Plateau shows an increasing trend, with a fitted slope of 0.01021 and an R2 of 0.96466. The Geodetector indicates that the factors with the greatest impact on vegetation cover in the Loess Plateau are temperature, precipitation, soil type, and land use. The highest average vegetation coverage is achieved when the temperature is between −4.8 and 2 °C or 12 and 16 °C, precipitation is between 630.64 and 935.51 mm, the soil type is leaching soil, and the land use type is forest. And the interaction between all factors has a greater effect on the vegetation cover than any single factor alone. This study reveals the factors influencing vegetation growth on the Loess Plateau, as well as their types and ranges, providing a scientific basis and guidance for improving vegetation coverage in this region.

The Digital Elevation Model (DEM) products acquired through traditional approaches, due to the limited spatial resolution, have challenges in representing terrace terrain. Although the Digital Surface Models (DSMs) obtained by Unmanned Aerial Vehicles (UAV) can reflect the finer structural elements (such as flat terrace beds and steep terrace risers) and local variation of terraces, they are often influenced by vegetation. Therefore, in this study, an efficient approach was proposed for constructing a DEM by integrating the GF-7 DEM with the DSM obtained via UAV during the fallow season. The newly generated DEM can present the detailed micro-topography and structural elements of terraces, while also effectively minimizing the interference of vegetation.

The soil erodibility factor (K) is the main data required for regional soil erosion investigation and mapping using soil erosion models. USLE-K, RUSLE2-K, EPIC-K and Dg-K are four widely used methods for calculating soil erodibility factor (K). However, it remains to be studied which algorithm is more suitable to calculate soil erodibility factor (K) in the global scale. While, soil erodibility factor (K) is mostly calculated based on soil physical and chemical property data, which does not involve the content of rock fragments in these algorithms. However, the amount of rock fragments and thier distribution difference have a certain influence on soil physical and chemical properties, and then affect the accuracy of the estimation of soil erodibility factor (K). In this paper, USLE-K, RUSLE2-K, EPIC-K and Dg-K algorithms were used to estimate global soil erodibility factor (K), and its spatial pattern and main controlling factors were analyzed. In this paper, the measured data of soil erodibility factor (K) were retrieved by literature search, and the measured database of K factor value was established. The rationality of the results of the above four algorithms was analyzed, and the above four algorithms for calculating K factor were modified according to the measured database of K factor. At the same time, USLE-K and RUSLE2-K algorithm are taken as an example to calculate the effect of rock fragments in the soil profile and rock fragments on the soil surface. The results showed that (1) The spatial pattern of global K factors estimated by the USLE-K, RUSLE2-K, EPIC-K and Dg-K models is similar, but the values in the K surfaces are different in some extent. (2) Comparing to 106 measured values, the mean value of estimated RUSLE2-K is the closest to the measured K factor, followed by the USLE-K algorithm and the EPIC-K algorithm, while the estimated K by Dg-K algorithm is quite different from the measured K factor. (3) The presence of rock fragment in the soil profile increased the global soil erodibility factor. The rock fragment on the soil surface reduces soil erodibility. This article made the calculation of K more complete and accurate, thereby improving the accuracy of regional soil erosion estimation. And provide the necessary scientific basis for the selection of K algorithms globally.

Gully erosion represents one of the most severe forms of land degradation. In regional management decisions, gully density serves as a crucial metric. As an essential measure under China’s national strategy for protecting black soil, gully erosion control projects rely on accurate simulation of gully density at the regional scale to enable more efficient and precise management. Taking the northeast China’s Songnen typical black soil region as the study area, this research employed a stratified unequal probability systematic sampling method to select 977 small watershed sample units. Using sub-meter resolution Google Earth imagery, gully density on farmland was visually interpreted. To ensure the interpretation’s accuracy, 55 typical small watersheds were randomly selected as validation units for field investigations. On this basis, the RF algorithm was used with 13 selected factors to predict farmland gully length density. The results showed the following: 1) The Random Forest model demonstrated high accuracy and applicability, with an NSE exceeding 0.5. Residuals primarily centered around -0.1 km/km². 2) Slope was identified as the key influencing factor for farmland gully density, followed by multi-year average May precipitation, slope length, and multi-year average rainstorm volume. Threshold analysis revealed a significant increase in gully density when slope exceeded 1.21° and slope length surpassed 74.15 m, but a weakening effect was observed when the slope and slope length reached certain thresholds. 3) The prediction results indicated higher gully densities in low mountains and hilly areas. Regions with a density range of 0–0.05 km/km², followed by 0.05–0.25 km/km². As density increased, the proportion of area gradually declined, with areas >1 km/km² accounting for no more than 15% of the total. High-density regions were concentrated in low mountains and hilly areas, with average gully densities of 1.33 km/km² and 1.80 km/km², respectively, whereas low-density regions were concentrated in plains, with densities close to 0 km/km². This study provides theoretical and technical support for regional gully management decisions in the black soil region of Northeast China, contributing to the protection of black soil resources. Keywords: Permanent gully; Songnen typical black soil region; Random Forest; Regional scale; Gully length density

Gully development is a significant geomorphological and environmental process that affects land degradation worldwide, with ephemeral gullies (EGs) and permanent gullies (PGs) being the two most common types. These two gully types are often spatially connected, and with such EG-PG connectivity can accelerate erosion. However, systematic research on this phenomenon remains limited, particularly at the regional scale. This study focuses on the spatial connectivity between EGs and PGs in the Songnen black soil region of northeast China. An unequal probability stratified sampling was used to establish 977 small watershed units, and a database of gullies and their connectivity was constructed based on sub-meter imagery. Among them, 55 representative units were randomly selected within geomorphic zones for field surveys and UAV validation to ensure data accuracy. Spatial patterns of gully connectivity were analyzed, and dominant controlling factors were identified using the Geodetector, which quantifies spatial stratified heterogeneity and evaluates the explanatory power of potential driving factors. The results are as follows: (1) Gully connectivity varies significantly across the region, with hotspot areas where more than 50% of permanent gullies are connected to ephemeral gullies, and cold spot clusters elsewhere. (2) Permanent gullies connected to ephemeral gullies differ significantly from unconnected ones in both length and width, with the former exhibiting a more elongated morphology. (3) Slope length and mean annual precipitation are the primary drivers of gully connectivity, both showing significant positive effects. Moreover, the interaction between mean annual precipitation and slope length shows the strongest explanatory power, indicating that precipitation, in combination with topographic features, plays a dominant role in shaping gully connectivity. By examining the spatial patterns of gully connectivity, this study contributes to a more refined understanding of gully morphological evolution and offers empirical insights for enhancing gully erosion models and optimizing regional soil and water conservation strategies.

Abstract: Climate change and human activities are seriously affecting the intensity and extent of soil erosion in the Pan-Third Pole region (PTP), which covers an area of approximately 5.14 × 107 km2. Accurate assessment of soil wind and water erosion is crucial for controlling soil degradation. In this study, soil water erosion in the PTP was estimated for 2018 using sampling units and the China Soil Loss Equation (CSLE), and soil wind erosion in the PTP from 1982 to 2020 was simulated using the Revised Soil Wind Erosion Equation (RWEQ), based on meteorological, soils, topographic, and remote sensing data. The results showed that: (1) Soil water erosion in the PTP mainly occurs in East Asia, South Asia, and the Black Sea coastal region, and the average soil wind erosion rate of the whole region is 263.4 t•km-2•a-1, and the average water erosion rate of the key erosion areas with water erosion rates exceeding 2,500 t•km-2•a-1 is 22.6 times higher than the average water erosion rate of the study area, and annual erosion amounted to 57.1×108t, accounting for 38.6% of total erosion amount. The soil water erosion rates of cropland, grassland, and forest were 525.7 t•km-2•a-1, 362.6 t•km-2•a-1, and 185.6 t•km-2•a-1, respectively. (2) Soil wind erosion in the PTP mainly occurs in cropland and grassland in semi-arid areas, and aeolian sand activity primarily occurring in extremely arid and arid areas (deserts), and the average multi-year soil wind erosion rate in regions other than deserts is 633.65 t•km-2•a-1, of which the mean soil wind erosion rate in the area where soil wind erosion rate is greater than 50 t•km-2•a-1 was 4,316.94 t•km-2•a-1, for cropland, grassland, and scrubland were 1,981.14 t•km-2•a-1, 3,815.05 t•km-2•a-1, and 4,010.95 t•km-2•a-1, respectively. (3) From 1982 to 2020, the soil wind erosion rate in the PTP decreased by 10.61 t•km-2•a-1. The proportion of the area with a decreasing trend was 19.53%, while the proportion of the area with an increasing trend was 28.35%. (4) Soil wind and water combined erosion mainly occur in cross-border regions of northern Syria, the Indus River Plain, the northern border of Iran and Afghanistan, the southwestern part of the Qinghai-Tibet Plateau, central Mongolia, the central part of the Loess Plateau, Inner Mongolia, and the bordering areas of the three eastern provinces, the average soil erosion rate of is 4,534.77 t•km-2•a-1, with the average soil erosion rates for grassland and cropland being 4,752.41 t•km-2•a-1 and 1,495.68 t•km-2•a-1, respectively. This study provided a comprehensive understanding of soil erosion (both soil wind and water erosion) in the PTP, and offered valuable data and decision-making support for current and future soil erosion prevention and ecological restoration projects.

Soil erosion is a global and interconnected biogeochemical process whose mechanisms are being profoundly altered by the interactions between climate change and vegetation dynamics. A critical scientific question is addressed in this study: how the effectiveness of vegetation restoration in reducing soil erosion is gradually undermined by increasingly frequent and intense extreme rainfall events. We integrate the Chinese Soil Loss Equation (CSLE) with high-resolution (30 m) satellite data and sub-hourly rainfall observations using the Google Earth Engine platform. Our nationwide assessment reveals an average soil erosion rate of 15.54 t ha-1 yr-1, with significant reductions in key erosion-prone regions: the Qinghai-Tibet Plateau (-8.14 %), Sichuan Hilly Basin (-16.17 %), Loess Plateau (-21.53 %), and Northeast China's Black Soil Region (-10.34 %). Vegetation recovery contributed to a long-term decline in erosion rates at 0.071 t ha-1 yr-1, yet this was partially offset by a 2.9 % increase in erosion linked to intensifying extreme rainfall events. Our findings demonstrate that climate-driven precipitation extremes are undermining gains from ecological restoration, highlighting the urgent need to incorporate extreme event resilience into global soil conservation strategies.

Topography plays a critical role in soil migration and redistribution, but few studies have been conducted to quantify its effects on sediment deposition. In this study, we established a physical simulation and analysis framework to investigate the erosional and depositional impacts of two-dimensional slope terrain, specifically applied to Xiannangou small watershed in the loess hilly region. The results showed that the slope gradient, slope length, and slope shape have significantly influence the distribution of soil erosion and deposition. The magnitude of the erosion/deposition rate (Yr) determines the relative intensity of slope erosion and deposition, where Yr < 0 indicates erosion and Yr > 0 signifies deposition. The erosion rate on straight slope exhibited a positive correlation with slope length, while the erosion/deposition rate on concave and convex slopes exhibited fluctuations with slope length. Soil erosion predominantly occurred along the main flow line and the middle slope, aligning with the observed distribution of gully and slope erosion in the field. The sediment deposition was primarily concentrated on the lower slope or the lowest outlet of the basin, notably in gullies and gentle slopes where the terrain slows, especially during transition from steep to gradual slopes. These results can effectively predict the relative erosion/sedimentation rate of two-dimensional slopes, significantly contributing to a comprehensive understanding of how topography influences soil erosion and deposition. This study thoroughly considers the role of sediment deposition in the soil erosion process, providing a more accurate reflection of soil erosion/deposition in small watersheds. It addresses the existing deficiency in sediment consideration within soil erosion evaluation and supports the enhancement of soil erosion model.

The Shuttle Radar Topography Mission (SRTM) is a digital representation of the terrain surface morphology that contains rich terrain information and is widely used in environmental analyses. However, SRTM is adversely affected by mixed noise, which typically include random and stripe noise. Mixed noise results in the significant loss of topographic information, which reduce the validity of related research. To eliminate mixed noise in SRTM data, we propose an adaptive low-rank group sparse model based on edge preservation (ALGS_EP) to remove mixed noise from datasets. The method relies on a low-rank group sparse model that considers the gradient features of the terrain. It calculates a terrain factor to adapt the noise elimination model to terrain changes. Additionally, it integrates with the edge structure of elevation data and applies a double-gradient constraint to preserve the structural details of the elevation data. The proposed model, built upon the alternating direction multiplier method framework, enhances the traditional weighted kernel paradigm minimization algorithm by introducing variable weights that adjust according to the gradient of elevation data during iterations. Additionally, it incorporates the correlation between strip noise and residual data blocks when computing the iteration count, ensuring an iterative solution approach that converges to the optimal solution. We used ALGS_EP to process global SRTM 1 data and published a higher-quality and higher-precision elevation dataset. The elevation data noise before and after noise elimination were statistically analyzed. Simulated and empirical results show that the model is highly robust and more effective than existing methods in both visual and quantitative evaluations. The noise elimination rate was 97.6%, compared to the original data. Therefore, this research was valuable for applications that use digital elevation model as an important data layer. This study proposes an adaptive group sparse noise elimination [(adaptive low-rank group sparse model based on edge preservation (ALGS_EP)] method based on edge preservation for addressing mixed noise in Shuttle Radar Topography Mission (SRTM). ALGS_EP effectively eliminates mixed noise while retaining terrain edge information more comprehensively. Compared to ICESat/Geoscience Laser Altimeter System (GLAS), the processed data shows improved vertical accuracy across different terrains. Additionally, the method successfully eliminates 97.6% of the noise in the global SRTM dataset. image

Topography is an important factor affecting soil erosion and is measured as a combination of the slope length and slope steepness (LS-factor) in erosion models, like the Chinese Soil Loss Equation. However, global high-resolution LS-factor datasets have rarely been published. Challenges arise when attempting to extract the LS-factor on a global scale. Furthermore, existing LS-factor estimation methods necessitate projecting data from a spherical trapezoidal grid to a planar rectangle, resulting in grid size errors and high time complexity. Here, we present a global 1-arcsec resolution LS-factor dataset (DS-LS-GS1) with an improved method for estimating the LS-factor without projection conversion (LS-WPC), and we integrate it into a software tool (LS-TOOL). Validation of the Himmelblau–Orlandini mathematical surface shows that errors are less than 1%. We assess the LS-WPC method on 20 regions encompassing 5 landform types, and R2 of LS-factor are 0.82, 0.82, 0.83, 0.83, and 0.84. Moreover, the computational efficiency can be enhanced by up to 25.52%. DS-LS-GS1 can be used as high-quality input data for global soil erosion assessment.

The Loess Plateau is one of the most severely affected regions by soil erosion in the world, with a fragile ecological environment. Vegetation plays a key role in the region's ecological restoration and protection. Most existing studies focus on the trends of vegetation cover change, while fewer studies investigate the driving factors or only conduct quantitative analyses. This study uses the Geodetector (Geographic Detector Model) to assess the impact of natural and human factors such as temperature, precipitation, soil type, and land use on vegetation growth, revealing the characteristics and driving mechanisms of vegetation cover changes on the Loess Plateau from 1982 to 2022. It also quantitatively calculates the influence of different factors. The findings indicate that vegetation cover on the Loess Plateau has generally risen from 1982 to 2022, but there was a noticeable difference around 2000. The annual mean slope of vegetation cover from 1982 to 2000 was 0.00129, which increased to 0.01075 from 2000 to 2022. The Geodetector indicates that the factors with the greatest impact on vegetation cover in the Loess Plateau are temperature, precipitation, soil type, and land use, and the interaction between these factors has a greater effect on vegetation cover than any single factor alone. This study identifies the ideal ranges or types for vegetation growth on the Loess Plateau over four time periods, offering a scientific basis for soil restoration and soil and water conservation efforts in the area.

Ephemeral gully (EG) erosion is an important type of water erosion. Understanding the spatial distribution of EGs and other influencing factors at a regional scale is crucial for developing effective soil and water management strategies. Unfortunately, this area has not been sufficiently studied. The present study visually interpreted the EGs based on Google Earth images in 137 small watersheds uniformly distributed in the Loess Plateau, compared them with measured results, and analyzed the factors influencing EG formation and density using GeoDetector. The results showed that visually interpreting EGs from Google Earth images was suitable for EG regional studies. Out of the 137 small watersheds, 33.6% had EG occurrence with an average density of 3.41 km/km2. Rainfall (R) and slope gradient (S) were the primary factors influencing the formation of EGs, while the area proportion of sloping farmland (APSF) and soil erodibility (K) were the main factors affecting EG density. The interaction of dual factors had a greater influence compared to single factors, with the interaction between S and Normalized Difference Vegetation Index (NDVI) having the greatest impact on EG formation and the interaction between K and NDVI on EG density. Although natural forces significantly influence whether EGs can form in a specific area, human activities greatly affect the density of the gullies that develop. This underscores the importance of proper land management in controlling gully erosion. These findings could provide theoretical support for EG prediction models and a scientific basis for soil and water loss control strategies at the regional scale.

The soil erodibility factor (K) is the main data required for regional soil erosion investigation and mapping using soil erosion models. Fine mapping of K and the study of the applicability of different K estimation methods at the global scale are important to improve the accuracy of global soil erosion evaluation. In this study, the USLE-K, RUSLE2-K, EPIC-K and Dg-K algorithms were used to calculate and compare the global K, and a global measured K database was established by using literature backtracking method and retrieval tool method. Spatial pattern and applicability analysis were carried out on the results of the above four algorithms. The four algorithms were corrected according to the measured K database. The results showed that (1) the global K spatial patterns obtained by the four algorithms were similar but slightly different, with the result of RUSLE2-K being the closest to the measured K, followed by the USLE-K and the EPIC-K, and the result of Dg-K differing significantly from the measured K. (2) The global K distribution characteristics showed some regularity with soil properties, such as soil silt content and sand content, with silt content having the greatest influence on K. (3) The results calculated by the corrected RUSLE2-K and USLE-K algorithms could meet the model applicability conditions and coincided with the results of local K mapping. The results of K mapping in this study on a global scale and the results of the comparative analysis of the applicability of different algorithms provide the necessary scientific basis for the selection of K algorithms globally and quantitative evaluation of soil erosion.