Prediction of phosphorus concentrations in shallow groundwater in intensive agricultural regions based on machine learning.

Excessive accumulation of phosphorus in soil profiles has become the main source of phosphorus in groundwater due to the application of phosphorus fertilizers in intensive agricultural regions (IARs). Elevated phosphorus concentrations in groundwater have become a global phenomenon, which places enormous pressure on the safe use of water resources and the safety of the aquatic environment. Currently, the prediction of pollutant concentrations in groundwater mainly focuses on nitrate nitrogen, while research on phosphorus prediction is limited. Taking the IARs approximately 8 plateau lakes in the Yunnan-Guizhou Plateau as an example, 570 shallow groundwater samples and 28 predictor variables were collected and measured, and a machine learning approach was used to predict phosphorus concentrations in groundwater. The performance of three machine learning algorithms and different sets of variables for predicting phosphorus concentrations in shallow groundwater was evaluated. The results showed that after all variables were introduced into the model, the R2, RMSE and MAE of support vector machine (SVM), random forest (RF) and neural network (NN) were 0.52-0.60, 0.101-0.108 and 0.074-0.081, respectively. Among them, the SVM model had the best prediction effect. The clay content and water-soluble phosphorus in soil and soluble organic carbon in groundwater had a high contribution to the prediction accuracy of the model. The prediction accuracy of the model with reduced number of variables showed that when the number of variables was equal to 6, the RF model had R2, RMSE and MAE values of 0.53, 0.108 and 0.074, respectively, and the number of variables increased again; there were small changes in R2, RMSE and MAE. Compared with the SVM and NN models, the RF model can achieve higher accuracy by inputting fewer variables.

Shallow groundwater fluctuation: An ignored soil N loss pathway from cropland.

Nitrogen (N) pollution originating from agricultural land is among the major threats to shallow groundwater (SG). Soil N losses due to the SG table fluctuation are neglected, although a large number of studies have been conducted to evaluate N losses through leaching and runoff. Herein, the characteristics of N losses driven by SG table fluctuation were investigated using the microcosm experiment and surveyed data from the croplands around Erhai Lake. According to the results achieved, the total N (TN) loss mainly occurred during the initial 12 days when the soil was flooded, then presented N immobilized by soil and finally, basically balanced between influent and effluent after 50 days. The results demonstrated that 1.7% of the original soil TN storage (0-100 cm) was lost. The alternation of drying and flooding could greatly increase TN loss up to 1086 kg hm-2, which was 2.72 times as much as that of continuous flooding flow. The amount of soil N losses to groundwater was closely related to the soil profile biochemical characteristics (water content, soil microbial immobilization, mineralization, nitrification, and denitrification processes). Soil N loss from crop fields driven by SG table fluctuation is 26 and 6 times of the runoff and leaching losses, respectively, while the soil N loss from the vegetable fields is 33 and 4 times of the runoff and leaching losses. The total amount of N losses from the croplands around the Erhai Lake caused by flooding of shallow groundwater (SG) in 2016 was estimated at 3506 Mg. The estimations showed that N losses would decrease by 16% if vegetables are replaced with staple food crops. These results imply that the adjustment of the planting structure was the key measure to reduce soil N storage and mitigate groundwater contamination.

Shift of lakeshore cropland to buffer zones greatly reduced nitrogen loss from the soil profile caused by the interaction of lake water and shallow groundwater.

The interaction of lake water (LW) and shallow groundwater (SGW) accelerates nitrogen (N) loss from the soil profile in the lakeshore cropland, and cropland buffer zone (CBZ) significantly inhibits N loss in this area. Here, characteristics of N loss and transformations driven by SGW and LW interactions were explored using microcosmic experiments, and N loss was estimated using in situ monitoring data before and after the construction of the CBZ along the west bank of Erhai Lake. The results indicated that NO3--N, dissolved organic N and total dissolved N sustained the main N losses in the soil, and the organic N was responsible for the main N loss in the effluent. The lower total nitrogen (TN) concentrations of SGW in this area, the greater the soil N loss. Moreover, N total loss from the 100 cm soil profile in the control check was 1.8 times that in the simulated SGW treatment. We found that nitrification, denitrification and anammox driven by the microbial community and N functional genes were the key processes leading to N loss. The effluent N (3.64%) and gaseous N (0.32%) loss ratios in the cropland for continuously growing vegetables (CGV) were much higher than that in the CBZ (1.07% of effluent N and 0.25% of gaseous N loss ratios). If a 100 m wide and 48 km long area of lakeshore cropland is CGV, an increase by 47% is projected by 2030 compared with the N loss in 2020. But this region was built as a 100 m wide CBZ or 50 m wide CBZ + 50 m wide CGV after 2019, N loss will be reduced by 87% and 44% in 2030 compared with the N loss in CGV. The results implied that restoring a suitable width of CBZ can significantly reduce N loss.

Electrical conductivity and dissolved oxygen as predictors of nitrate concentrations in shallow groundwater in Erhai Lake region.

Elevated nitrogen (N) concentration in shallow groundwater is becoming increasingly problematic, putting water resources under pressure. For more effective management of such a resource, more precise predictors of N level in groundwater using smart monitoring networks are needed. However, external factors such as land use type, rainfall, and N loads from multiple sources (residential and agricultural) make it difficult to accurately predict the spatial and temporal variations of N concentration. In order to identify the key factors affecting spatial and temporal N concentration in shallow groundwater and develop a predictive model, 635 groundwater samples from drinking wells in residential areas and agricultural wells in croplands of a typical agricultural watershed in the Erhai Lake Basin, southwest China, in the period from 2018 to 2020, were collected and analyzed. The results showed that the type of land use and seasonal variations significantly affected the N forms and their concentrations in the shallow groundwater, as the ratios of ON and NO3--N to TN were 30%-39% and 52%-59% for the two land uses and 25%-44% and 46%-66% for seasonal changes. Their variations were reflected by electrical conductivity (EC) and redox environment. EC and dissolved oxygen (DO) had a positive non-linear relationship with the concentrations of total nitrogen (TN) and nitrate (NO3--N). The fitted non-linear quantitative models were established separately to predict TN and NO3--N concentrations in groundwater using easily available indictors (EC and DO). The high accuracy and performance of the models were investigated and approved by rRMSE, MAE, and 1:1 line. These findings can provide technical support for the rapid prediction and evaluation of N pollution in shallow groundwater through easily available indicators.

Synthesis of a novel arginine-modified starch resin and its adsorption of dye wastewater.

In this work, corn starch (St) was firstly grafted with polyacrylamide (PAM) to obtain StAM, which was subsequently immobilized with arginine to obtain a guanidine-containing starch-based resin, StAM-Arg. The synthesized products were characterized via Fourier transform infrared spectroscopy (FT-IR), 13C-NMR nuclear magnetic resonance (13C-NMR), scanning electron microscopy (SEM), X-ray diffraction (XRD), gel permeation chromatography (GPC), X-ray photoelectron spectroscopy (XPS) and thermogravimetric analysis (TGA). StAM-Arg exhibited a significantly enhanced adsorption capacity for acid fuchsin (AF), acid orange G (AOG), and acid blue 80 (AB80) compared with zeolite, diatomite, St and StAM, and it also exhibited broad-spectrum adsorption for different dyes. Weak acidic conditions were favorable for the resin to adsorb acid dyes. The decolorization rate (DR) by StAM-Arg for mixed wastewater reached 82.49%, which was higher than that of activated carbon (DR = 58.09%). StAM-Arg showed high resistance to microbial degradation, resulting in significantly improved structural stability for the resin. Its antibacterial rate (AR) for E. coli was up to 99.73%. After 7 days in simulated natural water, the weight loss ratio (WR) of StAM-Arg was 14.5%, which was much lower than that of St (WR = 66.53%). The introduced guanidine groups were considered to be the major reason for the observed improvements. Furthermore, the cationic guanidine could trap the acid dyes via ion-exchange reactions, while effectively inhibiting or eliminating the growth of bacteria on the adsorbent surface. The above advantages, including good dyestuff adsorption properties, high structural stability and prolonged service life, make StAM-Arg overcome the inherent drawbacks of the existing natural polymer adsorbents and have good application prospect in the treatment of textile wastewater.