Short-term Lake Erie algal bloom prediction by classification and regression models

Influence of climate extremes on long-term changes in cyanobacterial blooms in a eutrophic and shallow lake

Climate change and human activities have crucial effects on the variations in phytoplankton blooms in lakes worldwide. A record-breaking heatwave and drought event was reported in the middle and lower reaches of the Yangtze River during the summer of 2022, but only little is known about how cyanobacterial blooms in lakes respond to such climate extremes. Here, we utilized MODIS images to generate the area, occurrence, and initial blooming date (IBD) of cyanobacterial blooms in Lake Chaohu from 2000 to 2022. We found that the area and occurrence of cyanobacterial blooms were largely reduced. At the same time, the IBD was delayed in 2022 compared with the previous 20 years. The annual occurrence and mean area of cyanobacterial blooms in 2022 were 17 % and 23.1 km2, respectively, which were the lowest reported levels since the 21st century. The IBD in 2022 was four months late compared with the IBD in 2020. The high wind speed in spring delayed the spring blooms in 2022. The record-breaking heatwaves and drought from June to August reduced the blooms by influencing the growth of cyanobacteria and reducing the flow of nutrients from the watershed into the lake. This study highlights the compound impact of heatwave and drought climate events on reducing cyanobacterial blooms in a long-term period, enhancing additional understanding of the changes in phytoplankton blooms in lakes.

Machine learning screening of biomass precursors to prepare biomass carbon for organic wastewater purification: A review

In the past decades, the amount of biomass waste has continuously increased in human living environments, and it has attracted more and more attention. Biomass is regarded as the most high-quality and cost-effective precursor material for the preparation carbon of adsorbents and catalysts. The application of biomass carbon has extensively explored. The efficient application of biomass carbon in organic wastewater purification were reviewed. With briefly introducing biomass types, the latest progress of Machine learning in guiding the preparation and application of biomass carbon was emphasized. The key factors in constructing efficient biomass carbon for adsorption and catalytic applications were discussed. Based on the functional groups, rich pore structure and active site of biomass carbon, it exhibits high efficiency in water purification performance in the fields of adsorption and catalysis. In addition, out of a firm belief in the enormous potential of biomass carbon, the remaining challenges and future research directions were discussed.

Machine learning framework for predicting cytotoxicity and identifying toxicity drivers of disinfection byproducts

Drinking water disinfection can result in the formation disinfection byproducts (DBPs, > 700 have been identified to date), many of them are reportedly cytotoxic, genotoxic, or developmentally toxic. Analyzing the toxicity levels of these contaminants experimentally is challenging, however, a predictive model could rapidly and effectively assess their toxicity. In this study, machine learning models were developed to predict DBP cytotoxicity based on their chemical information and exposure experiments. The Random Forest model achieved the best performance (coefficient of determination of 0.62 and root mean square error of 0.63) among all the algorithms screened. Also, the results of a probabilistic model demonstrated reliable model predictions. According to the model interpretation, halogen atoms are the most prominent features for DBP cytotoxicity compared to other chemical substructures. The presence of iodine and bromine is associated with increased cytotoxicity levels, while the presence of chlorine is linked to a reduction in cytotoxicity levels. Other factors including chemical substructures (CC, N, CN, and 6-member ring), cell line, and exposure duration can significantly affect the cytotoxicity of DBPs. The similarity calculation indicated that the model has a large applicability domain and can provide reliable predictions for DBPs with unknown cytotoxicity. Finally, this study showed the effectiveness of data augmentation in the scenario of data scarcity.

Employing hybrid deep learning for near-real-time forecasts of sensor-based algal parameters in a Microcystis bloom-dominated lake

Harmful cyanobacterial blooms (CyanoHABs) are increasingly impacting the ecosystem of lakes, reservoirs and estuaries globally. The integration of real-time monitoring and deep learning technology has opened up new horizons for early warnings of CyanoHABs. However, unlike traditional methods such as pigment quantification or microscopy counting, the high-frequency data from in-situ fluorometric sensors display unpredictable fluctuations and variability, posing a challenge for predictive models to discern underlying trends within the time-series sequence. This study introduces a hybrid framework for near-real-time CyanoHABs predictions in a cyanobacterium Microcystis-dominated lake - Lake Dianchi, China. The proposed model was validated using hourly Chlorophyll-a (Chl a) concentrations and algal cell densities. Our results demonstrate that applying decomposition-based singular spectrum analysis (SSA) significantly enhances the prediction accuracy of subsequent CyanoHABs models, particularly in the case of temporal convolutional network (TCN). Comparative experiments revealed that the SSA-TCN model outperforms other SSA-based deep learning models for predicting Chl a (R2 = 0.45-0.93, RMSE = 2.29-5.89 μg/L) and algal cell density (R2 = 0.63-0.89, RMSE = 9489.39-16,015.37 cells/mL) at one to four steps ahead predictions. The forecast of bloom intensities achieved a remarkable accuracy of 98.56 % and an average precision rate of 94.04 % ± 0.05 %. In addition, scenarios involving various input combinations of environmental factors demonstrated that water temperature emerged as the most effective driver for CyanoHABs predictions, with a mean RMSE of 2.94 ± 0.12 μg/L, MAE of 1.55 ± 0.09 μg/L, and R2 of 0.83 ± 0.01. Overall, the newly developed approach underscores the potential of a well-designed hybrid deep-learning framework for accurately predicting sensor-based algal parameters. It offers novel perspectives for managing CyanoHABs through online monitoring and artificial intelligence in aquatic ecosystems.

Quantitative modelling reservoir microalgae proliferation in response to water-soluble anions and cations influx

Atmospheric precipitation deposits acid-forming substances into surface water. However, the effects of water-soluble components on microalgae proliferation are poorly understood. This study analysed the growth characteristics of three microalgae bioindicators of water quality: Scenedesmus quadricauda, Chlorella vulgaris, and Scenedesmus obliquus, adopting on-site monitoring, culture experiments simulating 96 types of water by supplementing anions and cations, and predictive modelling. The result quantified pH > 3.0 rain with dominant Ca2+, Mg2+, and K+ cations, together with anions of NO3- and SO42-. The presence of Ca2+ of up to 0.1 mM and Mg2+ concentrations (>0.5 mM) suppressed Scenedesmus quadricauda growth. Soluble ions, luminosity, and pH had significant impacts (p ≤ 0.01) on increased microalgae proliferation. A newly proposed microalgae growth model predicted a 10.7-fold increase in cell density six days post-incubation in the case of rainfall. The modelling supports algal outbreaks and delays prediction during regional water cycles.

Elucidating the impacts of microplastics on soil greenhouse gas emissions through automatic machine learning frameworks

With the rise in global plastic production and agricultural demand, the released microplastics (MPs) have increasingly influenced the elemental cycles of soils, leading to notable effects on greenhouse gas emissions. Despite initial research, there remains a gap in establishing a detailed modeling approach that comprehensively explores the impacts of MPs on GHG emissions. Herein, we utilized literature mining to assemble a comprehensive dataset examining the interplays between MPs and emissions of CO2, CH4, and N2O. Five automated machine learning frameworks were employed for predictive modeling. The GAMA framework was particularly effective in predicting CO2 (Q2 = 0.946) and CH4 (Q2 = 0.991) emissions. The Autogluon framework provided the most accurate prediction for N2O emission, though it exhibited signs of overfitting. Interpretability analysis indicated that the type of MPs significantly influenced CO2 emission. Degradable MPs (i.e., polyamide) inherently led to elevated CO2 emission, and the environmental aging further exacerbated this effect. Although both linear and nonlinear correlations between MPs and CH₄ emission were not identified, the incorporation of specific MPs that elevate soil pH, augment soil water retention, and cultivate anaerobic conditions may potentially elevate soil CH₄ emission. This research underscores the profound influence of MPs on soil GHG emissions, providing vital insights for shaping agricultural policies and soil management practices in the context of escalating plastic use.

One to twelve-month-ahead forecasting of MODIS-derived Qinghai Lake area, using neuro-fuzzy system hybridized by firefly optimization

Lakes, as the main sources of surface water, are of great environmental and ecological importance and largely affect the climatic conditions of the surrounding areas. Lake area fluctuations are very effective on plant and animal biodiversity in the areas covered. Hence, accurate and reliable forecasts of ​lake area might provide the awareness of water and climate resources and the survival of various species dependent on area fluctuations. Using machine learning methods, the current study numerically predicted area fluctuations of ​China's largest lake, Qinghai, over 1 to 12 months ahead of lead time. To this end, Moderate Resolution Imaging Spectroradiometer (MODIS) sensor images were used to monitor the monthly changes in the area of ​the lake from 2000 to 2021. Predictive inputs included the MODIS-derived lake area time latency specified by the autocorrelation function. The data was divided into two periods of the train (initial 75%) and test (final 25%), and the input combinations were arranged so that the model in the test period could be used to predict 12 scenarios, including forecast horizons for the next 1 to 12 months. The adaptive neuro-fuzzy inference system (ANFIS) was utilized as a predictive model. The firefly algorithm (FA) was also used to optimize ANFIS and improve its accuracy, as a hybrid model ANFIS-FA. Based on evaluation criteria such as root mean square error (RMSE) (477-594 km2) and R2 (88-92%), the results confirmed the acceptable accuracy of the models in all forecast horizons, even long-term horizons (10 months, 11 months, and 12 months). Based on the normalized RMSE criterion (0.095-0.125), the models' performance was reported to be appropriate. Furthermore, the firefly algorithm improved the prediction accuracy of the ANFIS model by an average of 16.9%. In the inter-month survey, the models had fewer forecast errors in the dry months (February-March) than in the wet months (October-November). Using the current method can provide remarkable information about the future state of lakes, which is very important for managers and planners of water resources, environment, and natural ecosystems. According to the results, the current approach is satisfactory in predicting MODIS-derived fluctuations of Qinghai Lake area and has research value for other lakes.

Efficient smartphone-based measurement of phosphorus in water

Agricultural runoff is one of the main sources of excess phosphorus (P) in different water bodies, subsequently leading to eutrophication and harmful algal blooms. To effectively monitor P levels in water, there is a need for simple measurement tools and extensive public involvement to enable regular and widespread sampling. Several smartphone-based P measurement methods have been reported, which extract red-green-blue (RGB) values from colorimetric reactions to build statistical regression models for P quantification. However, these methods typically require meticulous light conditions, involve initial equipment investment, and have undergone limited testing for large-scale applications. To overcome these limitations, this study developed a smartphone-based, equipment-free and facile P colorimetric analysis method. Following the standard procedure of the ascorbic acid approach, colorimetric reactions were captured by a smartphone camera, and RGB values were extracted using Python code for modeling. Different indoor light conditions, phone types, containers, and types of water samples were examined, resulting in a collection of 1922 images. The best regression model, employing random forest with RGB values and container types as inputs, achieved an R2 of 0.97 and an RMSE of 0.051 for P concentrations ranging from 0.01 to 1.0 mg P/L. Additionally, the optimal classification model could estimate the level of P below 0.1 mg P/L with an accuracy of 95.2 (or 77.4 % for <0.05 mg P/L). The strong performance of the developed models, which are also available freely online, suggests an easy and effective tool for more frequent P measurement and greater public involvement.

Improved eutrophication model with flow velocity-influence function and application for algal bloom control in a reservoir in East China

Improving hydrodynamic conditions is considered an effective method for facilitating the eutrophication management. However, the effect of hydrodynamic conditions on algal growth has rarely been quantified. In this work, a eutrophication model was developed and flow velocity was introduced into the algae growth kinetic formula to simulate the dynamics of algae growth in a drinking water source reservoir in East China. Based on the previous research and model calibration, the flow velocity-influence function f(v) and its parameters were determined. Accordingly, the optimal flow velocity for the dominant algae growth and critical flow velocity for algal growth inhibition were presented to be 0.055 m/s and 0.200 m/s for the study reservoir. Modeled results considering f(v) agreed with better with observations and reproduced the algal overgrowth process more accurately. The spatial-temporal differences in chlorophyll a (Chl a) concentration distribution during the algal proliferation period were analyzed on the basis of simulation results, which corroborated the significant influence of flow velocity on algal growth. The established model was applied to investigate the effect of improvement in hydrodynamic conditions on algal bloom control in the reservoir, and the scenario simulation of the additional sluice was conducted. Results showed that the additional sluice operation inhibited algal overgrowth effectively, resulting in an average decrease of 24.8%, 3.3%, 43.0%, and 37.5% in modeled Chl a concentration upstream north, upstream south, midstream and downstream, respectively. The established model might serve as a practical tool for eutrophication management in the study reservoir and other water bodies with similar hydrological characteristics and geographical features.

Phosphate removal by low-cost industrial byproduct iron shavings: Efficacy and longevity

Iron shavings (IS) are low-cost industrial byproducts that show great potential in removing phosphorus (P) from contaminated water. This work investigates the effectiveness of IS for P (PO4-P) removal and emphasizes its pretreatment and longevity. A 4-d pretreatment of IS with 2.5 % NaCl resulted in a significant increase in P adsorption capacity, from approximately 1.0 to 2.5 mg/g. In column tests, the P removal efficiency remained above 60 % over 60 d, with a capacity of 4.1 mg P/g. Longevity tests involved seven adsorption-regeneration cycles, with an effective IS regeneration by 1 N NaOH and neutralization by HCl solution (pH=2), and the P adsorption capacity only slightly decreased from 2.14 to 1.75 mg P/g. To significantly improve the IS regeneration operation, we employed induction heating and compared an intermittent 10-s induction heating with an isothermal hot NaOH (85 ℃) treatment in 10-min desorption tests (95.3 % versus 56.6 % regeneration). We further found that IH completely regenerated IS in 5 min with 100 s of IH application, but 30 min were needed for hot NaOH (85 ℃) treatment. SEM/EDX, XRD, and XPS tests were conducted to track the changes in the morphology, crystallinity, and surface oxidation products of IS in the cycle tests. Notably, IS surface changed from coarse to smooth with fewer reactive sites and a higher conversion of amorphous Fe oxides to more crystalline Fe3O4, resulting in lower reactivity and fewer exposed Fe0 sites over multiple cycles. All of these mechanisms contributed to the deterioration in P removal capacity. Overall, this study provides a solid foundation for applying low-cost IS in effectively removing P from agricultural runoff.