Plateau zokors (Eospalax baileyi) are an agent actively involved in causing soil erosion and meadow degradation in the Yellow River Source Zone of West China. This study aims to quantify the amount of soil and nutrient loss from zokor mounds in relation to slope gradient and rainfall intensity, and to assess the amount of soil loss in zokor-infested areas compared with healthy meadows in Henan County, Qinghai Province. The results showed that zokor mounds were gradually lowered at a rate of 1.8–3.9 cm h⁻¹. Soil loss occurred two min after the rain began, reaching the maximum level during the first 20 min. The rate of soil loss and nutrient loss increased with the rainfall intensity and slope gradient. When the rainfall intensity rose from 5 to 10 mm h⁻¹, and from 10 to 15 mm h⁻¹, the total soil loss on 10° slopes increased by 2.5 times and 3.9 times, respectively, and soil nutrient loss increased by 1.7 times and 2.7 times, respectively. As the slope gradient steepened to 20°, the corresponding figures were 2.8 times and 4.3 times for total soil loss, and 1.8 times and 2.9 times, respectively, for soil nutrient loss. When the slope rose to 30°, the soil loss increased by 3.0 and 4.5 times, and the soil nutrient loss increased by 1.8 times and 3.1 times, respectively. There was a power function between soil loss and surface runoff (S = 0.2371Q^2.2307, R² = 0.9529). The soil was eroded at a rate of 256.6 g m⁻² h⁻¹ from zokor mounds, 17.7 times higher than in intact meadows, and 1.8 times higher than in partially recovered meadows. Most of the eroded soils had a mean diameter of 0–1.2 mm. It is recommended that artificial control of plateau zokors should be implemented, together with other ecological restoration measures to restrain the soil erosion problem caused by zokor activities.

The outer banks of meadow-type meandering river bends in the source zone of the Yellow River are especially vulnerable to bank failure. This study aims to understand how vegetation affects bank stability and the mechanism of bank failure, especially via a prediction of the width of a collapsed block of small rivers through a proposed bank stability equilibrium as well as field sampling. Soil and vegetation properties were surveyed at four sites near the riverbank in 2013–2016. It was found that the failed blocks had, on average, a dimension of 0.865 m (width) by 0.817 m (thickness) by 2.228 m (length). The variability in the size of all the failed blocks was attributed predominantly to the roots of plants. Block thickness could be logarithmically predicted by root length at R² ≥ 0.76. The block width predicted from the proposed equilibrium equation deviated from in situ measurements by approximately 22.1%, a discrepancy highly subject to the overestimation of root reinforcement using Wu’s model. By reducing the coefficient of Wu’s model from 1.2 to 0.85, the proposed equilibrium equation was reliable to predict the width of bank collapse. However, its applicability to other study areas needs to be verified in further studies.

Soil moisture plays an important role in maintaining ecosystem stability and sustainable development, especially for the upper reaches of the Yellow River region. Therefore, accurately and conveniently monitoring soil moisture has become the focus of scholars. This study combines three machine learning algorithms: random forest (RF), support vector machine (SVM), and back propagation neural network (BPNN)—with the traditional monitoring of soil moisture using remote sensing indices to construct a more accurate soil moisture inversion model. To enhance the accuracy of the soil moisture inversion model, 27 environmental variables were screened and grouped, including vegetation index, salinity index, and surface temperature, to determine the optimal combination of variables. The results show that screening the optimal independent variables in the Xijitan landslide distribution area lowered the root mean square error (RMSE) of the RF model by 16.95%. Of the constructed models, the combined model shows the best applicability, with the highest R² of 0.916 and the lowest RMSE of 0.877% with the test dataset; the further research shows that the BPNN model achieved higher overall accuracy than the other two individual models, with the test set R² being 0.809 and the RMSE 0.875%. The results of this study can provide a theoretical reference for the effective use of Landsat satellite data to monitor the spatial and temporal distribution of and change in soil water content on the two sides of the upper Yellow River basin under vegetation cover.

The plateau pika (Ochotona curzoniae) actively contributes to soil erosion and meadow degradation in western China’s Yellow River source zone. This study aimed to elucidate the effects of the pika mound numbers on the hydrodynamic characteristics and soil water erosion through simulated rainfall experiments. The inhibition effects of restored vegetation growth on meadow degradation and soil erosion were explored using a revegetated pika mound as a control. The results showed that at a rainfall intensity of 30 mm/h, the soil loss per unit time increased and then decreased with rainfall time and that 15–20 min of rainfall duration was the sensitive period for soil loss in different pika mound patch lands. The degradation of meadows due to the activities of plateau pika is an essential factor influencing soil erosion, and the soil erosion rate is positively correlated with both the pika mound numbers and slope. The mean flow velocity can better describe the process of the soil erosion, and its value increased with the number of pika mounds and slope. The Reynolds number ranged from 57.85 to 153.63 (Re <500), and it was preliminarily determined that all slope runoff was laminar flow. The Froude number increased linear function with pika mound numbers (p < 0.01), and both the pika mound numbers and slope were significant factors affecting it (p < 0.05). The Darcy–Weisbach resistance coefficient instead decreased with the pika mound numbers and slope, and the inhibitory effect of vegetation on soil erosion was probably limited when the mound numbers reached a certain level. According to the results of the grey correlation and Pearson correlation analysis, changes in the number of pika mounds led to variability in the soil water erosion properties by altering the landscape scale effect. The number of pika mound patches (NP), edge length index (TE), area index (AREA), and volume index (V) were the key influencing factors on the soil erosion rate and hydrodynamic parameters. We conclude that plateau pika activities intensify meadow degradation, and the continuous increase in pika mounds decreases the vegetation cover and flow resistance and increases soil erodibility. Controlling the intensity of plateau pika activities will effectively prevent soil erosion in degraded areas.