Interpretable machine learning guided by physical mechanisms reveals drivers of runoff under dynamic land use changes.

Human activities continuously impact water balances and cycling in watersheds, making it essential to accurately identify the responses of runoff to dynamic changes in land use types. Although machine learning models demonstrate promise in capturing the intricate interplay between hydrological factors, their "black box" nature makes it challenging to identify the dynamic drivers of runoff. To overcome this challenge, we employed an interpretable machine learning method to inversely deduce the dynamic determinants within hydrological processes. In this study, we analyzed land use changes in the Ningxia section of the middle Yellow River across four periods, laying the foundation for revealing how these changes affect runoff. The sub-watershed attributes and meteorological characteristics generated by the Soil and Water Assessment Tool (SWAT) model were used as input variables of the Extreme Gradient Boosting (XGBoost) model to simulate substantial sub-watershed rainfall runoff in the region. The XGBoost was interpreted using the SHapley Additive exPlanations (SHAP) to identify the dynamic responses of runoff to the land use changes over different periods. The results revealed increasingly frequent interchanges between the land use types in the study area. The XGBoost effectively captured the characteristics of the hydrological processes in the SWAT-derived sub-watersheds. The SHAP analysis results demonstrated that the promoting effect of agricultural land (AGRL) on runoff gradually weakens, while forests (FRST) continuously strengthen their restraining effect on runoff. Relevant land use policies provide empirical support for these findings. Furthermore, the interaction between meteorological variables and land use impacts the runoff generation mechanism and exhibits a threshold effect, with the thresholds for relative humidity (RH), maximum temperature (MaxT), and minimum temperature (MinT) determined to be 0.8, 25 °C, and 15 °C, respectively. This reverse deduction method can reveal hydrological patterns and the mechanisms of interaction between variables, helping to effectively addressing constantly changing human activities and meteorological conditions.

Pedo-transfer functions of the soil water characteristic curves of the vadose zone in a typical alluvial plain area in the lower reaches of the Yellow River using machine learning methods.

The soil water characteristic curve (SWCC) is of great significance for studying the hydrological cycle, agricultural water management, and unsaturated soil mechanics. However, it is difficult to effectively obtain a large number of SWCCs because of the cumbersome and expensive determination experiments for SWCCs. Pedo-transfer functions (PTFs) established using soil physicochemical properties have become an effective method for solving this problem. However, due to the limitations of the establishment methods and the wide spatial variability of soil properties, it is still difficult to establish PTFs in a specific region. In order to establish the PTFs of SWCCs for the alluvial plain area of the lower reaches of the Yellow River, 233 soil samples were collected from the vadose zone in a typical area. These data were used as the data sources, and eight variables including clay, silt content, fractal dimension, bulk density, total porosity, pH value, organic matter content, and electrical conductivity were used as the influencing factors. By applying and comparing three machine learning algorithms, the PTFs of the SWCCs based on the random forest algorithm were obtained. Based on the Gini index of the random forest, the insensitive factors were eliminated and the optimal variable input mode was constructed. Based on the verification, there was little difference between the predicted water content and the measured water content. The determination coefficient R2 is 0.9308; the root mean square error (RMSE) is 0.0447; and the mean relative error (MRE) is 22.40%.

Three-dimensional forward modeling and quantitative assessment of electrode offset effects in ERT.

Resistivity data has important applications in geophysical exploration, but the impact of electrode offsets on resistivity response characteristics remains unclear. This study aims to explore the influence of horizontal electrode offset angles and vertical offsets caused by topographical variations on the forward modeling of resistivity data. By analyzing experimental models with different measurement arrays, the paper revealed their influence laws on the buried depth of the target body and resistivity resolution. Utilizing tools like ZondRes3D, we conducted 3D resistivity forward modeling and analyzed the results in detail. It is found that horizontal electrode offsets lead to pseudo-anomalies in the apparent resistivity response, which is related to the offset angles and the number of electrodes. Under different conditions, the horizontal electrode offsets exhibit a "gradient variation" pattern. In addition, topographical variations can also cause distortions and offsets in the apparent resistivity curves and the locations of the anomaly response. Specifically, the measuring lines near the edge of the target bodies are more susceptible to these effects. Based on the comprehensive experimental results, we have drawn several conclusions regarding the impact of electrode offsets and topographical variations, including the effects of offset angles on the pseudo-anomalies, the anomalous response laws under different topographic conditions, as well as anomalous situations under specific angles. These findings provide crucial insights for interpreting resistivity data in geophysical exploration and addressing practical engineering problems, and offer guidance for optimizing measuring line layouts and post-processing terrain correction algorithms.

LNAPL migration processes based on time-lapse electrical resistivity tomography.

Contamination from light non-aqueous phase liquids (LNAPLs) and their derivatives, arising from exploration, production, and transportation, has become a prevalent pollution source. This poses direct threats to human health. However, conventional investigative methods face limitations when applied to studying the extent and migration process of LNAPL contamination, as well as the redistribution of LNAPL during groundwater level fluctuations. Conventional methods lack the ability to rapidly, efficiently, and in real-time acquire information about contaminated areas. Therefore, this study utilizes time-lapse electrical resistivity tomography to investigate the migration mechanism of LNAPL under unsaturated conditions, constant groundwater levels, and groundwater level reductions. A relationship between resistivity and water and oil contents was established and used for inverse calculation of LNAPL content via resistivity inversion. Time-lapse electrical resistivity tomography revealed LNAPL migration in a "concave" shape across three conditions. Groundwater presence notably slowed migration, hindering downward movement and leading to a floating oil band. A robust mathematical model was established to derive the relationship between resistivity and water and oil contents. Finally, LNAPL distribution under unsaturated conditions was inversely obtained from resistivity data, showing highest content at the top leak point, obstructed area, and bottom of soil column. Consequently, time-lapse electrical resistivity tomography demonstrates a notable capacity to characterize the LNAPL migration process. This technique constitutes an effective geophysical method for monitoring and describing the characteristics of LNAPL migration. Its significance lies in enhancing our understanding of remediation for LNAPL-induced groundwater and land contamination.

Multi-technological integration in a smelting site: Visualizing pollution characteristics and migration pattern.

Heavy metal(loid)s pollution of industrial legacies has become a severe environmental issue worldwide. Linking soil pollution to groundwater contaminant plumes would make invisible pollution features visible across the site, but related studies are lacking and require the convergence of multiple technologies. This study uniformly managed the soil and groundwater data in a 3D visualization model to pellucidly assess the spatial distribution of critical contaminants beyond simple drilling information. The distribution of Pb, Zn, As, and Cd in soil-groundwater system has a strong correlation to historical production, substance type, soil property, and groundwater flow direction. Over 2600 measurements of High-density electrical resistivity tomography (ERT) data were used to guarantee the exactness of soil structures. Hydraulic conductivity showed a strongest correlation (R2 = 0.86), yielding a calibrated model to reveal the anisotropic and contaminant transport in the region, with the consequent minimize the drilling tests. This study provides a template for the description of a verifiable scenario of hydrogeological conditions and pollution characteristics at smelting sites, coupled with traditional exploration and non-invasive techniques. The findings highlight the significance of visualizing the internal state of the soil-groundwater system under consideration, thus providing a basis for targeted control measures against site contamination.

Investigating Historical Baseflow Characteristics and Variations in the Upper Yellow River Basin, China.

The baseflow of the Yellow River is vital and important for water resource management and for understanding the hydrological cycle and ecohydrology setting in this arid and semi-arid basin. This study uses a Lyne and Hollick digital filtering technique to investigate the behaviors of the baseflow and the baseflow index in the upper reaches of the Yellow River Basin (China). The observed streamflow discharges along the river were used to analyze the baseflow trend, persistence, and periodic characteristics during the period of 1950-2000. The results show that the average baseflow and BFI in the upper reaches of the Yellow River exhibit a decreasing trend and will continue to decline in the future. Generally, the annual average baseflow and BFI for the most upstream areas of the Yellow River show little difference, while the baseflow and BFI exhibit significant differences for the downstream areas. The filtered annual baseflow varied between 128 × 108 m3/year and 193 × 108 m3/year for the Yellow River. The BFI ranged from 0.54 to 0.65, with an average of 0.60. This indicates that on average, 60% of the long-term streamflow is likely controlled by groundwater discharge and shallow subsurface flow. Statistics show that two periodic variations were observed in the baseflow evolution process. The results indicate that on average, the first and second main cycles of baseflow behaviors occur at 28 years and 12-17 years, respectively. Correspondingly, the estimation indicates that the abrupt change points tend to appear in the 1960s, the 1980s, and the 1990s. An improved understanding of baseflow behaviors can help guide future strategies to manage the river regime, its water resources, and water quality.

Effects of a Proposed Hydraulic Project on the Hydrodynamics in the Poyang Lake Floodplain System, China.

Knowledge of dam construction in floodplain systems and its hydrodynamic effects plays a critical role in managing various kinds of floodplains. This study uses 3D floodplain hydrodynamic modeling to explore the possible effects of a proposed hydraulic project in Poyang Lake (PLHP) on the hydrodynamics, exemplified by a large floodplain system. Simulations showed that the water levels across most lake regions presented more significant changes than in the floodplain areas during the study period. The increased water levels upstream from the PLHP (~1.0 m) were distinctly higher than that downstream (~0.1 m). The PLHP may decrease the magnitude of the water velocities in the main channels of the lake, whereas velocities may experience mostly minor changes in the floodplains, depending upon the altered flow dynamics and transport. On average, the water temperature may exhibit mostly minor changes (~<1.0 °C) for both the horizontal and vertical scales within the flood-pulse-influenced lake system. Additionally, the model results indicated that the outflow process caused by the PLHP may be altered from the natural discharge into the Yangtze River to frequent backflow events during the storage period, demonstrating the non-negligible effect of the PLHP on the water supply for the downstream Yangtze River in the future.