Which riverine water quality parameters can be predicted by meteorologically-driven deep learning?

River water quality has been significantly impacted by climate change and extreme weather events worldwide. Despite increasing studies on deep learning techniques for river water quality management, understanding which riverine water quality parameters can be well predicted by meteorologically-driven deep learning still requires further investigation. Here we explored the prediction performance of a traditional Recurrent Neural Network, a Long Short-Term Memory network (LSTM), and a Gated Recurrent Unit (GRU) using meteorological conditions as inputs in the Dahei River basin. We found that deep learning models (i.e., LSTM and GRU) demonstrated remarkable effectiveness in predicting multiple water quality parameters at daily scale, including water temperature, dissolved oxygen, electrical conductivity, chemical oxygen demand, ammonia nitrogen, total phosphorous, and total nitrogen, but not turbidity. The GRU model performed best with an average determination coefficient of 0.94. Compared to the daily-average prediction, the GRU model exhibited limited error increment of 10-40 % for most water quality parameters when predicting daily extreme values (i.e., the maximum and minimum). Moreover, deep learning showed superior performance in collective prediction for multiple water quality parameters than individual ones, enabling a more comprehensive understanding of the river water quality dynamics from meteorological data. This study holds the promise of applying meteorologically-driven deep learning techniques for water quality prediction to a broader range of watersheds, particularly in chemically ungauged areas.

Validation of the efficiency of arsenic mitigation strategies in southwestern region of Bangladesh and development of a cost-effective adsorbent to mitigate arsenic levels.

World's highest arsenic (As) contamination is well-documented for the groundwater system of southwestern region (mainly Jashore district) of Bangladesh, where the majority of inhabitants are underprivileged. To mitigate As poisoning in southwestern Bangladesh, numerous steps have been taken so far by the government and non-governmental organizations (NGOs). Among them, digging deep tube wells and As removal by naturally deposited Fe(OH)3 species are being widely practiced in the contaminated areas. However, these actions have been left unmonitored for decades, making people unaware of this naturally occurring deadly poison in their drinking water. Hence, water samples (n = 63, both treated and untreated) and soil samples (n = 4) were collected from different spots in Jashore district to assess the safety level of drinking water and to understand the probable reasons for high As(III) contamination. About 93.7% of samples were found to contain As(III) above 10 µg/L; among them, 38% contained above 50 µg/L. The study shows that current As(III) removal strategies in the study area are ineffective. In this connection, a simple low-cost As(III) removal adsorbent is proposed that can be prepared with very cheap and locally available materials like iron sludge and charcoal. The adsorbent was characterized in terms of SEM, EDX, and XPS. The optimal dosage of the adsorbent was investigated for real-life application concerning several vital water quality parameters. The Fe-C adsorbent exhibited a maximum As(III) removal efficiency of 92% in real groundwater samples. The study will allow policymakers for informed decision-making regarding water body management as well as enable the local people to avail As-safe water in a way that aligns with their economic factors.

Green metal oxides coated biochar nanocomposites preparation and its utilization in vertical flow constructed wetlands for reactive dye removal: Performance and kinetics studies.

Major causes of water pollution in the ecosystem are pollutants such as dyes which are noxious. The present study was based on the synthesis of the green nano-biochar composites from cornstalk and green metal oxide resulting in Copper oxide/biochar, Zinc oxide /biochar, Magnesium oxide/biochar, Manganese oxide/biochar, biochar for removal of dyes combined with the constructed wetland (CW). Biochar Augmentation in constructed wetland systems has improved dye removal efficiency to 95% in order of copper oxide/biochar > Magnesium oxide/biochar > Zinc oxide/biochar > Manganese oxide/biochar > biochar > control (without biochar) respectively in wetlands. It has increased the efficiency of pH by maintaining pH 6.9-7.4, while Total Suspended Solids (TSS) removal efficiency and Dissolved oxygen (DO) increased with the hydraulic retention time of about 7 days for 10 weeks. Chemical oxygen demand (COD) and colour removal efficiency increased with the hydraulic retention time of 12 days for 2 months and there was a low removal efficiency for total dissolved solids (TDS) from control (10.11%) to Copper oxide /biochar (64.44%) and Electrical conductivity (EC) from control (8%) to Copper oxide /biochar (68%) with the hydraulic retention time of about 7 days for 10 weeks. Colour and chemical oxygen demand removal kinetics followed second and first-order kinetic. A significant growth in the plants were also observed. These results proposed the use of agricultural waste-based biochar as part of a constructed wetland substratum can provide enhanced removal of textile dyes. That can be reused.

Total suspended solids characterization and management implications for lakes in East China.

In this study, we empirically developed a robust model (the Root Mean Square Error (RMSE), bias, NSE and RE were 26.63 mg/L, -4.86 mg/L, 0.47 and 16.47%, respectively) for estimating the total suspended solids (TSS) concentrations in lakes and reservoirs (Hereinafter referred to as lakes) across the Eastern Plain Lake (EPL) Zone. The model was based on 700 in-situ TSS samples collected during 2007-2020 and logarithmic transformed red band reflectance of Landsat data. Based on the Google Earth Engine (GEE), the TSS concentrations in 16,804 lakes were mapped from 1984 to 2019. The results demonstrated a decreasing tendency of TSS in 82.2% of the examined lakes (72.5% of the basins) indicating that the pollutants carried by TSS flowing into the lakes were decreasing. Statistically significant variation (p < 0.05) was found in half of these lakes (28.6% of the basins). High TSS level (>100 mg/L) was observed in 0.31% of lakes (1.1% of the basins). The changing rates of TSS in 47.8% of the lakes (52.7% of the basins) ranged between -50 mg/L/yr and 0. We found high and significantly increased relative spatial heterogeneity of TSS in 4.6% and 6.5% of lakes, respectively. Likewise, the environmental factors, i.e., fertilizer usage, domestic wastewater, industrial wastewater, precipitation, wind speed and Normalized Difference Vegetation Index (NDVI) exhibited a significant correlation with interannual TSS in 38, 21, 20, 11, 17 and 15 of the 91 basins, respectively. This analysis indicated that only precipitation and fertilizer usage were significantly (p < 0.05) related to the spatial distribution of TSS. The relative contributions of the six factors to the interannual TSS changes were varied in different basins. Overall, the NDVI (the representation of vegetation cover) had a high mean contribution to the interannual TSS changes with an average contribution of 7.2%, and contributions of fertilizer were varied greatly among the basins (0.01%-68%). Human activities (fertilizer usage, domestic wastewater, industrial wastewater) and natural factors (precipitation, wind speed and NDVI) played relatively important roles to TSS changes in 14 and 15 of the 91 basins, respectively. Beyond the six factors in this study, other unanalyzed factors (such as lake depth and soil texture) also had some impacts on the distribution of TSS in the study area.

Linking environmental with biological data: Low sampling frequencies of chemical pollutants and nutrients in rivers reduce the reliability of model results.

Linking environmental and biological data using ecological models can provide crucial knowledge about the effects of water quality parameters on freshwater ecosystems. However, a model can only be as reliable as its input data. Here, the influence of sampling frequency of temporal variable environmental input data on the reliability of model results when linked to biological data was investigated using Threshold Indicator Taxa Analysis (TITAN) and species sensitivity distributions (SSDs). Large-scale biological data from benthic macroinvertebrates and matching water quality data including four metals and four nutrients of up to 559 site-year combinations formed the initial data sets. To compare different sampling frequencies, the initial water quality data sets (n = 12 samples per year, set as reference) were subsampled (n = 10, 8, 6, 4, 2 and 1), annual mean values calculated and used as input data in the models. As expected, subsampling significantly reduced the reliability of the environmental input data across all eight substances. For TITAN, the use of environmental input data with a reduced reliability led to a considerable (1) loss of information because valid taxa were no longer identified, (2) gain of unreliable taxon-specific change points due to false positive taxa, and (3) bias in the change point estimation. In contrast, the reliability of the SSD results appeared to be much less reduced. However, closer examination of the SSD input data indicated that existing effects were masked by poor model performance. The results confirm that the sampling frequency of water quality data significantly influences the reliability of model results when linked with biological data. For studies limited to low sampling frequencies, the discussion provides recommendations on how to deal with low sampling frequencies of temporally variable water quality data when using them in TITAN, in SSDs, and in other ecological models.

Novel insights into impacts of the COVID-19 pandemic on aquatic environment of Beijing-Hangzhou Grand Canal in southern Jiangsu region.

In 2020, a sudden COVID-19 pandemic unprecedentedly weakened anthropogenic activities and as results minified the pollution discharge to aquatic environment. In this study, the impacts of the COVID-19 pandemic on aquatic environment of the southern Jiangsu (SJ) segment of Beijing-Hangzhou Grand Canal (SJ-BHGC) were explored. Fluorescent component similarity and high-performance size exclusion chromatography analyses indicated that the textile printing and dyeing wastewater might be one of the main pollution sources in SJ-BHGC. The water quality parameters and intensities of fluorescent components (WT-C1(20) and WT-C2(20)) decreased to low level due to the collective shutdown of all industries in SJ region during the Spring Festival holiday and the outbreak of the domestic COVID-19 pandemic in China (January 24th to late February, 2020). Then, they presented a gradual upward trend after the domestic epidemic was under control. In mid-March, the outbreak of the international COVID-19 pandemic hit the garment export trade of China and consequently inhibited the production activities of textile printing and dyeing industry (TPDI) in SJ region. After peaking on March 26th, the intensities of WT-C1(20) and WT-C2(20) decreased again with changed intensity ratio until April 12th. During the study period (135 days), correlation analysis revealed that WT-C1 and WT-C2 possessed homology and their fluorescence intensities were highly positively correlated with conductivity and CODMn. With fluorescence fingerprint (FF) technique, this study not only excavated the characteristics and pollution causes of water body in SJ-BHGC, but also provided novel insights into impacts of the COVID-19 pandemic on production activities of TPDI and aquatic environment of SJ-BHGC. The results of this study indicated that FF technique was an effective tool for precise supervision of water environment.

[Inversion of Water Quality Parameters Based on UAV Multispectral Images and the OPT-MPP Algorithm].

Unmanned aerial vehicle (UAV) multispectral remote sensing can be used to monitor multiple water quality parameters, such as suspended solids, turbidity, total phosphorus, and chlorophyll. Establishing a stable and accurate water quality parameter inversion model is a prerequisite for this work. The matching pixel-by-pixel (MPP) algorithm is an inversion algorithm for high resolution features of UAV images; however, it is associated with problems of excessive computation and over-fitting. To overcome these problems, the optimize-MPP (OPT-MPP) algorithm is proposed. In this study, Qingshan Lake in Hangzhou City, Zhejiang Province, was used as the research area. Forty-five samples were collected to construct the OPT-MPP algorithm inversion model for two water quality parameters:the suspended sediments concentration (SS) and turbidity (TU). The results showed that the optimal suspended sediment concentration inversion model had a determination coefficient (R2) of 0.7870 and a comprehensive error of 0.1308. The optimal turbidity inversion model had a R2 of 0.8043 and a comprehensive error of 0.1503. Hence, the inversion of the spatial distribution information for water quality parameters in each experimental area of QingShan Lake was realized by using the optimal models of the two established parameters.

Spatially Varying and Scale-Dependent Relationships of Land Use Types with Stream Water Quality.

Understanding the complex relationships between land use and stream water quality is critical for water pollution control and watershed management. This study aimed to investigate the relationship between land use types and water quality indicators at multiple spatial scales, namely, the watershed and riparian scales, using the ordinary least squares (OLS) and geographically weighted regression (GWR) models. GWR extended traditional regression models, such as OLS to address the spatial variations among variables. Our results indicated that the water quality indicators were significantly affected by agricultural and forested areas at both scales. We found that extensive agricultural land use had negative effects on water quality indicators, whereas, forested areas had positive effects on these indicators. The results also indicated that the watershed scale is effective for management and regulation of watershed land use, as the predictive power of the models is much greater at the watershed scale. The maps of estimated local parameters and local R2 in GWR models showcased the spatially varying relationships and indicated that the effects of land use on water quality varied over space. The results of this study reinforced the importance of watershed management in the planning, restoration, and management of stream water quality. It is also suggested that planners and managers may need to adopt different strategies, considering watershed characteristics-such as topographic features and meteorological conditions-and the source of pollutants, in managing stream water quality.

Real-time detection of potable-reclaimed water pipe cross-connection events by conventional water quality sensors using machine learning methods.

Risk of cross-connection is becoming higher due to greater construction of potable-reclaimed water dual distribution systems. Cross-connection events can result in serious health concerns and reduce public confidence in reclaimed water. Thus, reliable, cost-effective and real-time online detection methods for early warning are required. The current study carried out pilot-scale experiments to simulate potable-reclaimed water pipe cross-connection events for different mixing ratios (from 30% to 1%) using machine learning methods based on multiple conventional water quality parameters. The parameters included residual chlorine, pH, turbidity, temperature, conductivity, oxidation-reduction potential and chemical oxygen demand. The results showed that correlated variation occurred among water quality parameters at the time of the cross-connection event. A single parameter-based method can be effective at high mixing ratios, but not at low mixing ratios. The direct supporting vector machine (SVM)-based method managed to overcome this drawback, but coped poorly with abnormal readings of water parameter sensors. In that respect, a Pearson correlation coefficient (PCC)-SVM-based method was developed. It provided not only high detection performance under normal conditions, but also remained reliable when abnormal readings occurred. The detection accuracy and true positive rate of this method was still over 88%, and the false positive rate was below 12%, given a sudden variation of an individual water quality parameter. The receiver operating characteristic curves further confirmed the promising practical applicability of this PCC-SVM-based method for early detection of cross-connection events.

Holistic water quality dynamics in rural artificial shallow water bodies.

The water environment in diversely used rural artificial water bodies is generally varied with seasonal and diurnal changes, stability of which is significant in water resources management. Understanding of interaction among different water quality parameters that depend on their diurnal variations is the concern of this study. A rural homestead pond used for aquaculture in Bangladesh and a micro-dam used as an irrigation tank for paddy farming in Japan are chosen for contrasting the analysis of data. The observed data series of four typical water quality parameters exhibits the diurnal variations, which are primarily inferred to be driven by solar radiation and complex bio-chemical interactions. The study proposes a stochastic differential equation model to represent holistic water quality dynamics based on continuous measurements. The water quality parameters are considered as temporally continuous Markov process, where their individual effects on each parameter are evaluated in a specific time step and immediately reflected to the next observation. The model parameters are calibrated and the stability is discussed based on the eigenvalues of model parameters. The result mostly shows the mean reverting properties for dissolved oxygen and water temperature, while pH and oxidation reduction potential are rather depend on other parameters or external disturbance.

Multi-scale analysis of the relationship between landscape patterns and a water quality index (WQI) based on a stepwise linear regression (SLR) and geographically weighted regression (GWR) in the Ebinur Lake oasis.

Water quality is highly dependent on landscape characteristics. This study explored the relationships between landscape patterns and water quality in the Ebinur Lake oasis in China. The water quality index (WQI) has been used to identify threats to water quality and contribute to better water resource management. This study established the WQI and analyzed the influence of landscapes on the WQI based on a stepwise linear regression (SLR) model and geographically weighted regression (GWR) models. The results showed that the WQI was between 56.61 and 2886.51. The map of the WQI showed poor water quality. Both positive and negative relationships between certain land use and land cover (LULC) types and the WQI were observed for different buffers. This relationship is most significant for the 400-m buffer. There is a significant relationship between the water quality index and landscape index (i.e., PLAND, DIVISION, aggregation index (AI), COHESION, landscape shape index (LSI), and largest patch index (LPI)), demonstrated by using stepwise multiple linear regressions under the 400-m scale, which resulted in an adjusted R 2 between 0.63 and 0.88. The local R 2 between the LPI and LSI for forest grasslands and the WQI are high in the Akeqisu River and the Kuitun rivers and low in the Bortala River, with an R 2 ranging from 0.57 to 1.86. The local R 2 between the LSI for croplands and the WQI is 0.44. The local R 2 values between the LPI for saline lands and the WQI are high in the Jing River and low in the Bo River, Akeqisu River, and Kuitun rivers, ranging from 0.57 to 1.86.

Seasonal variability of estuarine dynamics due to freshwater discharge and its influence on biological productivity in Yeongsan River Estuary, Korea.

In order to evaluate water quality and biological productivity, observation data sets were collected and analyzed in Yeongsan River Estuary, Korea. We also set up a numerical model to resolve hydrodynamics and fate of water quality variables in the system. Results show that most of nutrients loading are trapped in the lake and higher concentrations of nutrients and organic matters (OM) are present only inside of the artificial sea dike. There exist episodial discharges at the dam, which coincide mostly with rainfall events during summer monsoon periods. During this discharge event, lower salinity and higher suspended solids, nutrients, and OM are observed in surface layer of the estuarine section. Hydrodynamic model results show that circulation in the estuarine section is governed by freshwater discharge from the lake, resulting in an enhanced two-layer estuarine circulation being dominated, during and after the freshwater is discharged. Such two-layer estuarine circulation combined with higher concentration of nutrients in the surface layer results in that outfluxes of nutrients in the surface layer dominate over the influxes in the bottom layer during summer high precipitation periods. Meanwhile, numerical dye experiment results show that the discharged water with elevated nutrients levels have a short residence time (∼5-10 days) in the estuarine section. Due to this fast flushing rate, excessive nutrient loadings are not used to produce biological matters in the estuarine section. This limited biological productivity, characterized by seaward side of the artificial sea dike, makes Yeongsan estuarine system excluded from acting as an active carbon sink.

Adsorption of microcystin-LR on mesoporous carbons and its potential use in drinking water source.

Microcystin-LR (MC-LR) is a common toxin derived from cyanobacterial blooms an effective, rapid and non-toxic method needs to be developed for its removal from drinking water treatment plants (DWTP). For an adsorption-based method, mesoporous carbon can be a promising supplemental adsorbent. The effect of mesoporous carbon (MC1, MC2, and MC3) properties and water quality parameters on the adsorption of MC-LR were investigated and the results were analyzed by kinetic, isotherm, thermodynamic, Derjaguin-Landau-Verwey-Overbeek (DLVO), and intraparticle diffusion models. MC1 was the most appropriate type for the removal of MC-LR with a maximum adsorption capacity of 35,670.49 µg/g. Adsorption of MC-LR is a spontaneous reaction dominated by van der Waals interactions. Pore sizes of 8.5-14 nm enhance the pore diffusion of MC-LR from the surface to the mesopores of MC1. The adsorption capacity was not sensitive to changes in the pH (3.2-8.0) and the existence of organic matter (2-5 mg/L). Furthermore, the final concentration of MC-LR was below the WHO guideline level after a 10-min reaction with 20 mg/L of MC1 in the Nak-Dong River, a drinking water source. The MC-LR adsorption mainly competed with humic substances (500-1000 g/mole); however, they did not have a great effect on adsorption.

[Absorption and Reflection Characteristics of Black Water Blooms in the Eutrophic Water].

Studying on optical properties of black water blooms, is the precondition for using remote sensing technology to monitor and evaluate the black water blooms event. Black water blooms occurred in Taihu Lake in July 2015. A total of 36 water samples were observed in the three water regions of Taihu Lake, the region 1 with black water blooms characteristics, region 2 with cyanobacterial bloom characteristics, and regional 3 with characteristics of clean lake water. The reflectance spectra and absorption coefficient of these three regions were analyzed, and the results show that:1 The absorption coefficients of the total particles, the pigment particles and the non-pigment particlesin black water blooms are 1 to 2 times higher than the other two areas. The absorption coefficient of CDOM between 400-500 nm in region 1 is about 2 times higher than the other two areas, which lead the black water area with a very lower reflectance, and presents as black color. 2 The range of M value in black water blooms is lower than Dianchi Lake and Chaohu Lake, which means the humic acid content of CDOM with black water characteristics is higher. A significant positive correlation is found between chlorophyll a (Chl-a) and the CDOM absorption coefficient at 350 nm, indicating that algae degradation is likely to be the primary source of CDOM in black waters. 3 The contribution of each optically active component indicates that the water absorption of region 1 strongly controlled by CDOM below 380 nm, but by Chla absorption between 380 nm and 700 nm.