Toxic cyanobacterial blooms have become a severe global hazard to human and environmental health. Most studies have focused on the relationships between cyanobacterial composition and cyanotoxins production. Yet, little is known about the environmental conditions influencing the hazard of cyanotoxins. Here, we analysed a unique 22 sites dataset comprising monthly observations of water quality, cyanobacterial genera, zooplankton assemblages, and microcystins (MCs) quota and concentrations in a large-shallow lake. Missing values of MCs were imputed using a non-negative latent factor (NLF) analysis, and the results achieved a promising accuracy. Furthermore, we used the Bayesian additive regression tree (BART) to quantify how Microcystis bloom toxicity responds to relevant physicochemical characteristics and zooplankton assemblages. As expected, the BART model achieved better performance in Microcystis biomass and MCs concentration predictions than some comparative models, including random forest and multiple linear regression. The importance analysis via BART illustrated that the shade index was overall the best predictor of MCs concentrations, implying the predominant effects of light limitations on the MCs content of Microcystis. Variables of greatest significance to the toxicity of Microcystis also included pH and dissolved inorganic nitrogen. However, total phosphorus was found to be a strong predictor of the biomass of total Microcystis and toxic M. aeruginosa. Together with the partial dependence plot, results revealed the positive correlations between protozoa and Microcystis biomass. In contrast, copepods biomass may regulate the MC quota and concentrations. Overall, our observations arouse universal demands for machine-learning strategies to represent nonlinear relationships between harmful algal blooms and environmental covariates.

Water level prediction in large dammed rivers is an important task for flood control, hydropower generation, and ecological protection. The variations of water levels in large rivers are traditionally simulated based on hydrological models. Recently, most studies have begun applying deep learning (DL) models as an alternative method for forecasting the dynamics of water levels. However, it is still challenging to directly apply DL to the simultaneous prediction of water levels across multiple sites. This study attempts to develop a hybrid framework by combining the Physical-based Hydrological model (PHM) and Long Short-Term Memory (LSTM). This study hypothesizes that our hybrid model can enhance the predictive accuracy of water levels in large rivers, because it considers the temporal-spatial information of mainstream-tributaries relationships. The effectiveness of the proposed model (PHM-BP-LSTM) is evaluated using the daily water levels from 2012 to 2018 in the Three Gorges Reservoir (TGR), China. Firstly, we use a hydrological model to produce a large amount of water level data to solve the limited training data set. Then, we use the Back Propagation (BP) neural network to capture the mainstream-tributaries relationship. The future changes in water levels in the different mainstream stations are simultaneously predicted by the LSTM model. We reveal that our hybrid model yields satisfactory accuracy for daily water level simulations at fourteen mainstream stations of the TGR. We further demonstrate the proposed model outperforms the traditional machine learning methods in different prediction scenarios (one-day-ahead, three-day-ahead, seven-day-ahead), with RMSE values ranging from 0.793 m to 1.918 m, MAE values ranging from 0.489 m to 1.321 m, and the average relative errors at each mainstream station are controlled below 4%. Overall, our PHM-BP-LSTM, combining physical process and deep learning, can be viewed as a potentially useful approach for water level prediction in the TGR, and possibly for the rapid forecast of changes in water levels in other large rivers.

Harmful algal blooms (HABs) have been deteriorating global water bodies, and the accurate prediction of algal dynamics using the modelling method is a challenging research area. High-frequency monitoring and deep learning technology have opened up new horizons for HAB forecasting. However, the non-stationary and stochastic process behind algal dynamics monitoring largely limits the prediction performance and the early warning of algal booms. Through an analysis of the published literature, we found that decomposition methods are widely used in time-series analysis for hydrological processes. Predictions of ecological indicators have received less attention due to their inherent fluctuations. This study explores and demonstrates the predictive enhancement for chlorophyll fluorescence data based on the coupling of three decomposition algorithms with conventional deep learning models: the convolutional neural network (CNN) and long short-term memory (LSTM). We found that the decomposition algorithms can successfully capture the time-series patterns of chlorophyll fluorescence concentrations. The results indicate that decomposition-based models can enhance the accuracy of single models in predicting chlorophyll concentrations in terms of the improvement percentages in RMSE (with increases ranging from 25.7% to 71.3%), MAE (ranging from 28.3% to 75.7%), and R² values (increasing ranging from 14.8% to 34.8%). In addition, the comparison experiment for different decomposition methods might suggest the superiority of singular spectral analysis in hourly predictive tasks of chlorophyll fluorescence over the wavelet transform and empirical mode decomposition models. Overall, while decomposition methods come with their respective strengths and weaknesses, they are undeniably efficient in combination with deep learning models in dealing with the high-frequency monitoring of chlorophyll fluorescence data. We also suggest that model developers pay more attention to online data preprocessing and conduct comparative analyses to determine the best model combinations for forecasting algal blooms and water management.

Water quality data can supply a substantial decision support for water resources utilization and pollution prevention. However, there are numerous missing values in water quality data due to inescapable factors like sensor failure, thereby leading to biased result for hydrological analysis and failing to support environmental governance decision accurately. A Latent Factorization of Tensors (LFT) with Stochastic Gradient Descent (SGD) proves to be an efficient imputation method. However, a standard SGD-based LFT model commonly surfers from the slow convergence that impairs its efficiency. To tackle this issue, this paper proposes a Fast Latent Factorization of Tensors (FLFT) model. It constructs an adjusted instance error into SGD via leveraging a nonlinear PID controller to incorporates the past, current and future information of prediction error for improving convergence rate. Comparing with state-of-art models in real world datasets, the results of experiment indicate that the FLFT model achieves a better convergence rate and higher accuracy.

The gradual deterioration of water quality caused by human activities and global warming can lead to severe public health and economic consequences. Despite the critical need for effective water quality (WQ) monitoring using networks of automated sensors to mitigate these risks, the data collected are inevitably contaminated with missing values and insufficient monitoring frequency, which can hinder WQ prediction, assessment, and early warning. Various methods have been proposed to address the issue of incomplete data in follow-up applications through missing value imputation. However, the existing imputation approaches of incomplete WQ data fail to fully consider certain important mutual relationships among the spatio-temporal measurements and the imbalanced distribution of known data in an high-dimensional and incomplete (HDI) WQ tensor. Latent factorization of tensors (LFT) has been demonstrated to be particularly effective in many other applications. In this paper, we propose a Biased Non-negative Latent factorization of tensors with Nonlinear Regularization (BNLNR) model. This study adopts a nonlinear swish regularization scheme, aiming at enhancing the solution space of a resultant Non-negative Latent factorization of tensors (NLFT) model so as to improve its representative learning ability. Investigations using two large high-frequency water quality datasets reveal that the imputation accuracy of proposed method considerably outperforms several state-of-art baselines for different testing cases. Experimental results on two high-frequency WQ datasets from real applications demonstrate that compared with state-of-the-art models, the BNLNR model achieves prominent imputation accuracy for unobserved WQ data.