Agricultural, environmental and ecological modeling requires soil cation exchange capacity (CEC) that is difficult to measure. Pedotransfer functions (PTFs) are thus routinely applied to predict CEC from easily measured physicochemical properties (e.g., texture, soil organic matter, pH). This study developed the support vector machines (SVM)‐based PTFs to predict soil CEC based on 208 soil samples collected from A and B horizons in Qingdao City, Shandong Province, China. The database was randomly split into calibration and validation datasets in proportions of 3:1 using the bootstrap method. The optimal SVM parameters were searched by applying the genetic algorithm (GA). The performance of SVM models was compared to those of multiple stepwise regression (MSR) and artificial neural network (ANN) models. Results show that the accuracy of CEC predicted by SVM improves considerably over those predicted by MSR and ANN. The performance of SVM for B horizon (R2 = 0.85) is slightly better than that for A horizon (R2 = 0.81). The SVM is a powerful approach in the simulation of nonlinear relationship between CEC and physicochemical properties of widely distributed samples from different soil horizons. Sensitivity analysis was also conducted to explore the influence of each input parameter on the CEC predictions by SVM. The clay content is the most sensitive parameter, followed by soil organic matter and pH, while sand content has the weakest influence. This suggests that clay is the most important predictor for predicting CEC of both soil horizons.

In order to explore the water level variations of Caizi Lake under river-lake isolation, the monthly water level of the Chefuling station in Caizi Lake from 1989 to 2018 and the daily water level, rainfall and flow of local hydrological stations in 2018 were analyzed by using the Mann-Kendall trend test and wavelet analysis. Results showed that the difference of the average water level of Caizi Lake between the flood and dry seasons was 3.34 m, with a multi-year average water level of 10.42 m above sea level. The first and second main periods of the water level of Caizi Lake were 128 and 18 months, respectively, with 4 and 29 "up-down" cycles, respectively. From 2018, the next 3-4 years were likely to be the low water level period. The water level of Caizi Lake was significantly correlated with that of the Anqing hydrological station of the Yangtze River (r=0.824, P<0.01). In addition, the current hydrological staging of Caizi Lake was about 30 days behind than before the sluice was built. Under the dual influences of the river-lake isolation and the Yangtze-to-Huaihe Water Diversion Project (YHWD), the hydrological regime change of Caizi Lake and its eco-environmental effect needed long-term monitoring and research.

(1) Background: Simulation of soil water infiltration process and analysis of its influencing factors are important for water resources management. (2) Methods: In this study, the relative contributions of the soil water retention characteristics (SWRC) estimation, initial water content, and constant pressure head at upper boundary to the cumulative infiltration under various soil conditions were quantified based on the 1-D Richards’ equation and 900 scenarios. Scenario simulations were performed for two SWRC estimation methods (Jensen method and Rosetta); three different initial water contents (0.15, 0.20, and 0.25 cm3/cm3); five different constant pressure heads (0.5, 1, 2, 4, and 8 cm); and thirty soil samples with varying texture and bulk density. (3) Results: Rosetta representing the drying branch of the SWRC yielded higher simulated cumulative infiltration compared with the Jensen method representing the wetting branch of the SWRC. However, the Jensen method–predicted cumulative infiltration fluxes matched well with the measured values with a low RMSE of 0.80 cm. (4) Conclusions: The relative contribution of the SWRC estimation method to cumulative infiltration (19.1–72.2%) was compared to that of constant pressure head (14.0–65.5%), and generally greater than that of initial water content (2.2–29.9%). Findings of this study have practical significance for investigating the transport of water, nutrients, and contaminants in the unsaturated zone.

Abstract. Soil available water capacity (AWC) plays an important role in soil and water resource management, and is time-consuming and expensive to measure over large areas. AWC is often predicted with pedotransfer functions (PTFs) using easily measured soil properties (e.g., texture, bulk density and organic matter content). This study investigated the effectiveness of using neural network based-ROSETTA PTF to estimate AWC from four hierarchical levels of soil information: textural class (R1), texture (R2), texture and bulk density (R3), texture, bulk density, and water content at -33 kPa (θ -33 ) and -1500 kPa (θ -1500 ) (θ -33 and θ -1500 derived from Gupta-Larson PTF) (R4). A comprehensive dataset containing 58 soil samples collected from eastern China were used for evaluation. Results show that all the ROSETTA models except R1 estimate reliably the AWC. R2 and R3 produce similar estimates, both of which have only slightly poorer performance than R4. In this case soil texture is considered to be the most important soil property influencing AWC, making negligible the effects of the bulk density and organic matter content. It can be concluded that measuring soil texture as inputs for ROSETTA is the most affordable and reasonable approach to predict AWC in the study area.

Abstract. Global sensitivity analysis of the parameters of the modified universal soil loss equation (MUSLE) was conducted by using the extended Fourier amplitude sensitivity test (EFAST) method. Results show that the runoff factor, slope length and gradient factor, crop management factor and erosion control practice factor were the most sensitive parameters which affected the model outputs, followed by soil texture, organic matter content, aggregation class and the class of the water permeability of the soil profile. The number of iterations for Monte Carlo simulation had significant influence on the sensitivity analysis results. Only when the number of iterations was larger than 50000, EFAST can yield satisfactory convergence of sensitivity indices. Overall, the interaction of the runoff factor, soil erodible factor and adjustment factors was substantial when using the MUSLE. The key parameters should be prioritized for calibration to determine the most optimal values and reduce the uncertainty in soil loss predictions.