Chlorophyll a predictability and relative importance of factors governing lake phytoplankton at different timescales

Developing a real-time water quality simulation toolbox using machine learning and application programming interface

Rivers are vital for sustaining human life as they foster social development, provide drinking water, maintain aquatic ecosystems, and offer recreational spaces. However, most rivers are being increasingly contaminated by pollutants from non-point sources, urbanization, and other sources. Consequently, real-time river water quality modeling is essential for managing and protecting rivers from contamination, and its significance is growing across various sectors, including public health, agriculture, and water treatment systems. Therefore, a real-time river water quality simulation toolbox was developed using machine learning (ML) and an application program interface (API). To create the toolbox, models that simulated water quality parameters such as chlorophyll a (Chl-a), dissolved oxygen (DO), total nitrogen (TN), total organic carbon (TOC), and total phosphorus (TP) at each point in the Nakdong River were constructed. The models were constructed using Artificial neural network (ANN), Random Forest (RF), support vector machines (SVM), and data from API. Subsequently, hyperparameter optimization was conducted to enhance the model's performance. During training, the models' performances were evaluated and compared based on the data sampling method and ML algorithms. Models trained with random sampling data outperformed those trained with time-series data. Among the algorithm models that used random sampling data, the RF exhibited the best performance. The average coefficient of determination (R2) values for each water quality simulation with randomly sampled data using RF for DO, TN, TP, Chl-a, and TOC were 0.79, 0.65, 0.74, 0.45, and 0.48, respectively. For ANN, they were 0.7, 0.51, 0.64, 0.35, and 0.35, respectively, and for SVM, they were 0.73, 0.51, 0.59, 0.21, and 0.3, respectively. The Chl-a and TOC models exhibited relatively poor performance, whereas the DO, TN, and TP models demonstrated superior performance. Diversifying the input data variables is necessary to improve the performance of the Chl-a and TOC models. Sensitivity and uncertainty analyses were conducted to evaluate and enhance the models' understanding. Furthermore, using a graphic user interface (GUI) toolbox, user convenience was maximized.

Hyperspectral indices data fusion-based machine learning enhanced by MRMR algorithm for estimating maize chlorophyll content

Accurate estimation of chlorophyll is essential for monitoring maize health and growth, for which hyperspectral imaging provides rich data. In this context, this paper presents an innovative method to estimate maize chlorophyll by combining hyperspectral indices and advanced machine learning models. The methodology of this study focuses on the development of machine learning models using proprietary hyperspectral indices to estimate corn chlorophyll content. Six advanced machine learning models were used, including robust linear stepwise regression, support vector machines (SVM), fine Gaussian SVM, Matern 5/2 Gaussian stepwise regression, and three-layer neural network. The MRMR algorithm was integrated into the process to improve feature selection by identifying the most informative spectral bands, thereby reducing data redundancy and improving model performance. The results showed significant differences in the performance of the six machine learning models applied to chlorophyll estimation. Among the models, the Matern 5/2 Gaussian process regression model showed the highest prediction accuracy. The model achieved R2 = 0.71 for the training set, RMSE = 338.46 µg/g and MAE = 264.30 µg/g. In the case of the validation set, the Matern 5/2 Gaussian process regression model further improved its performance, reaching R2 =0.79, RMSE=296.37 µg/g, MAE=237.12 µg/g. These metrics show that Matern's 5/2 Gaussian process regression model combined with the MRMR algorithm to select optimal traits is highly effective in predicting corn chlorophyll content. This research has important implications for precision agriculture, particularly for real-time monitoring and management of crop health. Accurate estimation of chlorophyll allows farmers to take timely and targeted action.

The interaction between water quality and meteorological factors on chlorophyll-a concentration in Honghu Lake: based on PiecewiseSEM-generalized additive model coupling model

In order to explore the interactive effects of environmental factors on the chlorophyll-a (Chl-a) concentration variation in Honghu Lake, this study was based on the monitoring data of Chl-a mass concentration and water quality factors (water temperature, pH, dissolved oxygen, permanganate index, total nitrogen, total phosphorus) and meteorological factors (evaporation, precipitation, sunshine hours, average wind speed) at three research sites (Dakou, Chatan Island, Lantian) in Honghu Lake from January 2010 to December 2019. Time series analysis, piecewise structural equation model (PiecewiseSEM), and generalized additive model (GAM) were used to quantitatively study the spatial and temporal changes of different environmental factors and their interaction with chlorophyll-a concentration in Honghu Lake. The results showed that the effects of TN and DO on Chl-a at Dakou and Chatan Island were more significant than other environmental meteorological factors, while the effects of DO and CODMn on Chl-a at Lantian were more obvious. At the same time, it was found that Chl-a had a non-linear relationship with TN and DO at Dakou and Chatan Island, a non-linear relationship with DO at Lantian, and a linear relationship with CODMn. The interaction effect of dominant environmental meteorological factors on Chl-a was significantly higher than that of a single factor, and the explanation rates were 80.6%, 72.8%, and 64.6%, respectively. In conclusion, based on the Piecewise SEM and GAM model, it not only can reveal the influence of the interaction of influencing factors on the change of Chl-a concentration, but also has important significance for the early warning and control of lake eutrophication.

An Intelligent Early Warning System for Harmful Algal Blooms: Harnessing the Power of Big Data and Deep Learning

Harmful algal blooms (HABs) pose a significant ecological threat and economic detriment to freshwater environments. In order to develop an intelligent early warning system for HABs, big data and deep learning models were harnessed in this study. Data collection was achieved utilizing the vertical aquatic monitoring system (VAMS). Subsequently, the analysis and stratification of the vertical aquatic layer were conducted employing the "DeepDPM-Spectral Clustering" method. This approach drastically reduced the number of predictive models and enhanced the adaptability of the system. The Bloomformer-2 model was developed to conduct both single-step and multistep predictions of Chl-a, integrating the " Alert Level Framework" issued by the World Health Organization to accomplish early warning for HABs. The case study conducted in Taihu Lake revealed that during the winter of 2018, the water column could be partitioned into four clusters (Groups W1-W4), while in the summer of 2019, the water column could be partitioned into five clusters (Groups S1-S5). Moreover, in a subsequent predictive task, Bloomformer-2 exhibited superiority in performance across all clusters for both the winter of 2018 and the summer of 2019 (MAE: 0.175-0.394, MSE: 0.042-0.305, and MAPE: 0.228-2.279 for single-step prediction; MAE: 0.184-0.505, MSE: 0.101-0.378, and MAPE: 0.243-4.011 for multistep prediction). The prediction for the 3 days indicated that Group W1 was in a Level I alert state at all times. Conversely, Group S1 was mainly under an Level I alert, with seven specific time points escalating to a Level II alert. Furthermore, the end-to-end architecture of this system, coupled with the automation of its various processes, minimized human intervention, endowing it with intelligent characteristics. This research highlights the transformative potential of integrating big data and artificial intelligence in environmental management and emphasizes the importance of model interpretability in machine learning applications.

Assessing abundance-suitability models to prioritize conservation areas for the dwarf caimans in South America

Species-environment relationships have been extensively explored through species distribution models (SDM) and species abundance models (SAM), which have become key components to understand the spatial ecology and population dynamics directed at biodiversity conservation. Nonetheless, within the internal structure of species' ranges, habitat suitability and species abundance do not always show similar patterns, and using information derived from either SDM or SAM could be incomplete and mislead conservation efforts. We gauged support for the abundance-suitability relationship and used the combined information to prioritize the conservation of South American dwarf caimans (Paleosuchus palpebrosus and P. trigonatus). We used 7 environmental predictor sets (surface water, human impact, topography, precipitation, temperature, dynamic habitat indices, soil temperature), 2 regressions methods (Generalized Linear Models-GLM, Generalized Additive Models-GAM), and 4 parametric distributions (Binomial, Poisson, Negative binomial, Gamma) to develop distribution and abundance models. We used the best predictive models to define four categories (low, medium, high, very high) to plan species conservation. The best distribution and abundance models for both Paleosuchus species included a combination of all predictor sets, except for the best abundance model for P. trigonatus which incorporated only temperature, precipitation, surface water, human impact, and topography. We found non-consistent and low explanatory power of environmental suitability to predict abundance which aligns with previous studies relating SDM-SAM. We extracted the most relevant information from each optimal SDM and SAM and created a consensus model (2,790,583 km2) that we categorized as low (39.6%), medium (42.7%), high (14.9%), and very high (2.8%) conservation priorities. We identified 279,338 km2 where conservation must be critically prioritized and only 29% of these areas are under protection. We concluded that optimal models from correlative methods can be used to provide a systematic prioritization scheme to promote conservation and as surrogates to generate insights for quantifying ecological patterns.

Evaluating the relative importance of predictors in Generalized Additive Models using the gam.hp R package

Generalized Additive Models (GAMs) are widely employed in ecological research, serving as a powerful tool for ecologists to explore complex nonlinear relationships between a response variable and predictors. Nevertheless, evaluating the relative importance of predictors with concurvity (analogous to collinearity) on response variables in GAMs remains a challenge. To address this challenge, we developed an R package named gam.hp. gam.hp calculates individual R 2 values for predictors, based on the concept of 'average shared variance', a method previously introduced for multiple regression and canonical analyses. Through these individual R 2s, which add up to the overall R 2, researchers can evaluate the relative importance of each predictor within GAMs. We illustrate the utility of the gam.hp package by evaluating the relative importance of emission sources and meteorological factors in explaining ozone concentration variability in air quality data from London, UK. We believe that the gam.hp package will improve the interpretation of results obtained from GAMs.

Non-traditional abiotic drivers explain variability of chlorophyll-a in a shallow estuarine embayment

Understanding the factors influencing eutrophication, as represented by concentrations of chlorophyll-a (Chl-a), is needed to inform effective management and conservation strategies promoting ecological resilience. The objective of this study was to evaluate a unique combination of abiotic explanatory factors to describe Chl-a concentrations within the study estuary (North Biscayne Bay, Florida, USA). Multiple linear regression determined the strength and direction of influence of factors using data from 10 water quality monitoring stations. The analysis also considered time scales for evaluating cumulative effects of freshwater inflow and wind. Results show that dominant drivers of Chl-a were temperature, freshwater volume (whose cumulative effects were evaluated up to a 60-day time scale), and turbidity, which were statistically significant at 60, 60, and 70 % of the investigated stations, respectively. All drivers collectively accounted for 22 to 63 % of the variability of Chl-a measurements. Of the nine variables evaluated, nutrient concentrations (orthophosphate and ammonia) were not among the top three overall drivers. Despite nutrients historically being cited in the literature as the most significant factor, this study asserts that non-nutrient factors often govern Chl-a levels, necessitating a paradigm shift in management strategies to bolster estuarine resilience against climate change.

Integrating impacts of climate change on aquatic environments in inter-basin water regulation: Establishing a critical threshold for best management practices

Inter-basin water diversion (IBWD) is a viable strategy to tackle water scarcity and quality degradation due to climate change and increasing water demand in headwaters regions. Nevertheless, the capacity of IBWD to mitigate the impacts of climate change on water quality has rarely been quantified, and the underlying processes are not well understood. Therefore, this study aims to elucidate how the IBWD manipulated total phosphorus (TP) loading dilution and conveying patterns under climate change and determine a critical threshold for the quantity of water entering downstream reservoirs (WIN) for operational scheduling. To resolve this issue, climate-driven hydrologic variability over a 60-year period was derived utilizing the least square fitting approach. Subsequently, six scenarios evaluating the response of in-lake TP concentrations (TPL) to increased temperatures and IBWDs of 50 %, 100 %, and 150 % from the baseline water volume in 2030 and 2050 were studied by employing a calibrated hydrological-water quality model (SWAT-YRWQM). In the next stage, three datasets derived from mathematical statistics based on the observed data, the Vollenweider formula, and modeled projections were integrated to formulate best management practices. The results revealed that elevated air temperatures would lead to reduced annual catchment runoff but increased IBWD. Additionally, our study quantified the IBWD potential for mitigating water quality degradation, indicating the adverse effects of climate change on TPL would be weakened by 4.2-14.4 %. A critical threshold for WIN was also quantified at 617 million m3, maintaining WIN at or near 617 million m3 through optimized operational scheduling of IBWD could effectively restrict external inflow TP loading to lower levels. This study clearly illustrates the intricate interactive effects of climate change and IBWD on aquatic environments. The methodology elucidated in this study for determining the critical threshold of WIN could be applied in water management for analogous watershed-receiving waterbody systems.

Decision tree ensemble with Bayesian optimization to predict the spatial dynamics of chlorophyll-a concentration: A case study in Bay of Bengal

An accurate prediction of the spatial distribution of phytoplankton biomass, as represented by Chlorophyll-a (CHL-a) concentrations, is important for assessing ecological conditions in the marine environment. This study developed a hyperparameter-optimized decision tree-based machine learning (ML) models to predict the geographical distribution of marine phytoplankton CHL-a in the Bay of Bengal. To predict CHL-a over a large spatial extent, satellite-derived remotely sensed data of ocean color features (CHL-a, colored dissolved organic matter, photosynthetically active radiation, particulate organic carbon) and climatic factors (nighttime sea surface temperature, surface absorbed longwave radiation, sea level pressure) from 2003 to 2022 are used to train and test the models. Results obtained from this study have shown the highest concentrations of CHL-a occurred near the Bay's coastal belts and river estuaries. Analysis revealed that aside from photosynthetically active radiation, organic components exhibited a stronger positive relationship with CHL-a than climatic features, which are correlated negatively. Results showed the chosen decision tree methods to all possess higher R2 and lower root mean square error (RMSE) errors. Furthermore, XGBoost outperforms all other models in predicting the geographic distribution of CHL-a. To assess the model efficacy on seasonal basis, a best performing XGBoost model was validated in the Bay of Bengal region which has shown a good performance in predicting the spatial distribution of Chl-a as well as the pixel values during the summer, winter and monsoon seasons. This study provides the best ML model to researchers for predicting CHL-a in the Bay of Bengal. Further it helps to improve our knowledge of CHL-a spatial dynamics and assist in monitoring marine resources in the Bay of Bengal. It worth noting that the water quality in the Indian Ocean is very dynamic in nature, therefore, additional efforts are needed to test the efficacy of this study model over different seasons and spatial gradients.

Two machine learning approaches for predicting cyanobacteria abundance in aquaculture ponds

Cyanobacteria blooms in aquaculture ponds harm the harvesting of aquatic animals and threaten human health. Therefore, it is crucial to identify key drivers and develop methods to predict cyanobacteria blooms in aquaculture water management. In this study, we analyzed monitoring data from 331 aquaculture ponds in central China and developed two machine learning models - the least absolute shrinkage and selection operator (LASSO) regression model and the random forest (RF) model - to predict cyanobacterial abundance by identifying the key drivers. Simulation results demonstrated that both machine learning models are feasible for predicting cyanobacterial abundance in aquaculture ponds. The LASSO model (R² = 0.918, MSE = 0.354) outperformed the RF model (R² = 0.798, MSE = 0.875) in predicting cyanobacteria abundance. Farmers with well-equipped aquaculture ponds that have abundant water monitoring data can use the nine environmental variables identified by the LASSO model as an operational solution to accurately predict cyanobacteria abundance. For crude ponds with limited monitoring data, the three environmental variables identified by the RF model provide a convenient solution for useful cyanobacteria prediction. Our findings revealed that chemical oxygen demand (COD) and total organic carbon (TOC) were the two most important predictors in both models, indicating that organic carbon concentration had a close relationship with cyanobacteria growth and should be considered a key metric in water monitoring and pond management of these aquaculture ponds. We suggest that monitoring of organic carbon coupled with phosphorus reduction in feed usage can be an effective management approach for cyanobacteria prevention and to maintain a healthy ecological state in aquaculture ponds.

The impact of climate change on the water quality of Baiyangdian Lake (China) in the past 30 years (1991-2020)

Climate change significantly influenced the water quality of lakes in recent decades. This study investigated the effects of climate change on the water quality of Baiyangdian Lake (China) in the past 30 years (1991-2020) using correlation analysis, regression analysis, and the generalized additive model (GAM). The results show that water quality grade, chemical oxygen demand (COD), total phosphorus (TP) concentrations, and annual average and minimum air temperatures of the lake showed significant differences (p < 0.05) in the one-way ANOVA during the studied period. The concentration of dissolved oxygen (DO) and TP, annual average and minimum air temperatures, and annual precipitation decreased, while the COD and total nitrogen (TN) concentration, annual maximum temperature, and monthly maximum precipitation increased. The annual average and minimum air temperature affected all water quality variables and explained 12.3 %-54.5 % of variation deviation in correlation and GAM analyses, indicating that the changes of air temperature influenced the water temperature, which then affected the biochemical reaction rates leading to changes in water quality. The precipitation factors explained 10.5 % (TN) to 54.8 % (TP) of variation deviation, implying that the increase in precipitation improved water quality by diluting the COD concentration. However, excessive precipitation also accelerated the endogenous release of phosphorus in sediments by increasing the TP concentration. Additionally, extreme climate factors correlated with some water quality variables and explained 57.7 %-95.9 % of the total variances in correlation and regression analyses, suggesting that the extreme temperatures changed the nitrogen and DO concentration to aggravate lake pollution. However, the extreme precipitation purified the water through dilution. This study will facilitate to understand the impacts of climate change on water quality and find appropriate adaptation measures for ecosystem management of shallow lakes.

Large scale seasonal forecasting of peak season algae metrics in the Midwest and Northeast U.S

In recent decades, many inland lakes have seen an increase in the prevalence of potentially harmful algae. In many inland lakes, the peak season for algae abundance (summer and early fall in the northern hemisphere) coincides with the peak season for recreational use. Currently, little information regarding expected algae conditions is available prior to the peak season for productivity in inland lakes. Peak season algae conditions are influenced by an array of pre-season (spring and early summer) local and global scale variables; identifying these variables for forecast development may be useful in managing potential public health threats posed by harmful algae. Using the LAGOS-NE dataset, pre-season local and global drivers of peak-season algae metrics (represented by chlorophyll-a) are identified for 178 lakes across the Northeast and Midwest U.S. from readily available gridded datasets. Forecasting models are built for each lake conditioned on relevant pre-season predictors. Forecasts are assessed for the magnitude, severity, and duration of seasonal chlorophyll concentrations. Regions of pre-season sea surface temperature, and pre-season chlorophyll-a demonstrate the most predictive power for peak season algae metrics, and resulting models show significant skill. Based on categorical forecast metrics, more than 70% of magnitude models and 90% of duration models outperform climatology.  Forecasts of high and severe algae magnitude perform best in large mesotrophic and oligotrophic lakes, however, high algae duration performance appears less dependent on lake characteristics. The advance notice of elevated algae biomass provided by these models may allow lake managers to better prepare for challenges posed by algae during the high use season for inland lakes.

Hierarchical attention network for multivariate time series long-term forecasting

Multivariate time series long-term forecasting has always been the subject of research in various fields such as economics, finance, and traffic. In recent years, attention-based recurrent neural networks (RNNs) have received attention due to their ability of reducing error accumulation. However, the existing attention-based RNNs fail to eliminate the negative influence of irrelevant factors on prediction, and ignore the conflict between exogenous factors and target factor. To tackle these problems, we propose a novel Hierarchical Attention Network (HANet) for multivariate time series long-term forecasting. At first, HANet designs a factor-aware attention network (FAN) and uses it as the first component of the encoder. FAN weakens the negative impact of irrelevant exogenous factors on predictions by assigning small weights to them. Then HANet proposes a multi-modal fusion network (MFN) as the second component of the encoder. MFN employs a specially designed multi-modal fusion gate to adaptively select how much information about the expression of current time come from target and exogenous factors. Experiments on two real-world datasets reveal that HANet not only outperforms state-of-the-art methods, but also provides interpretability for prediction.

Evaluating the eutrophication risk of artificial lagoons-case study El Gouna, Egypt

Eutrophication problem in El Gouna shallow artificial coastal lagoons in Egypt was investigated using 2D TELEMAC-EUTRO-WAQTEL module. Eight reactive components were presented, among them dissolved oxygen (DO), phosphorus, nitrogen, and phytoplankton biomass (PHY). The effect of warmer surface water on the eutrophication problem was investigated. Also, the spatial and temporal variability of the eutrophication was analyzed considering different weather conditions: tide wave, different wind speeds and directions. Moreover, effect of pollution from a nearby desalination plant was discussed considering different pollution degrees of brine discharge, different discharge quantities and different weather conditions. Finally, new precautions for better water quality were discussed. The results show that tide wave created fluctuations in DO concentrations, while other water quality components were not highly influenced by tide's fluctuations. Also, it was found that high water temperatures and low wind speeds highly decreased water quality producing low DO concentrations and high nutrients rates. High water quality was produced beside inflow boundaries when compared to outflow boundaries in case of mean wind. Moreover, the results show that the average water quality was not highly deteriorated by the nearby desalination operation, while the area just beside the desalination inflow showed relatively strong effects. Different weather conditions controlled the brine's propagation inside the lagoons. Moreover, increasing the width of the inflow boundaries and injecting tracer during tide and mean wind condition are new precautions which may help to preserve the water quality in a future warmer world. This study is one of the first simulations for eutrophication in manmade lagoons.

Near-term phytoplankton forecasts reveal the effects of model time step and forecast horizon on predictability

As climate and land use increase the variability of many ecosystems, forecasts of ecological variables are needed to inform management and use of ecosystem services. In particular, forecasts of phytoplankton would be especially useful for drinking water management, as phytoplankton populations are exhibiting greater fluctuations due to human activities. While phytoplankton forecasts are increasing in number, many questions remain regarding the optimal model time step (the temporal frequency of the forecast model output), time horizon (the length of time into the future a prediction is made) for maximizing forecast performance, as well as what factors contribute to uncertainty in forecasts and their scalability among sites. To answer these questions, we developed near-term, iterative forecasts of phytoplankton 1-14 days into the future using forecast models with three different time steps (daily, weekly, fortnightly), that included a full uncertainty partitioning analysis at two drinking water reservoirs. We found that forecast accuracy varies with model time step and forecast horizon, and that forecast models can outperform null estimates under most conditions. Weekly and fortnightly forecasts consistently outperformed daily forecasts at 7-day and 14-day horizons, a trend that increased up to the 14-day forecast horizon. Importantly, our work suggests that forecast accuracy can be increased by matching the forecast model time step to the forecast horizon for which predictions are needed. We found that model process uncertainty was the primary source of uncertainty in our phytoplankton forecasts over the forecast period, but parameter uncertainty increased during phytoplankton blooms and when scaling the forecast model to a new site. Overall, our scalability analysis shows promising results that simple models can be transferred to produce forecasts at additional sites. Altogether, our study advances our understanding of how forecast model time step and forecast horizon influence the forecastability of phytoplankton dynamics in aquatic systems and adds to the growing body of work regarding the predictability of ecological systems broadly.

Uncovering the influence of hydrological and climate variables in chlorophyll-A concentration in tropical reservoirs with machine learning

Climate variability and change, associated with increasing water demands, can have significant implications for water availability. In the Brazilian semi-arid, eutrophication in reservoirs raises the risk of water scarcity. The reservoirs have also a high seasonal and annual variability of water level and volume, which can have important effects on chlorophyll-a concentration (Chla). Assessing the influence of climate and hydrological variability on phytoplankton growth can be important to find strategies to achieve water security in tropical regions with similar problems. This study explores the potential of machine learning models to predict Chla in reservoirs and to understand their relationship with hydrological and climate variables. The model is based mainly on satellite data, which makes the methodology useful for data-scarce regions. Tree-based ensemble methods had the best performances among six machine learning methods and one parametric model. This performance can be considered satisfactory as classical empirical relationships between Chla and phosphorus may not hold for tropical reservoirs. Water volume and the mix-layer depth are inversely related to Chla, while mean surface temperature, water level, and surface solar radiation have direct relationships with Chla. These findings provide insights on how seasonal climate prediction and reservoir operation might influence water quality in regions supplied by superficial reservoirs.

Chlorophyll a variations and responses to environmental stressors along hydrological connectivity gradients: Insights from a large floodplain lake

Understanding the key drivers of eutrophication in floodplain lakes has long been a challenge. In this study, the Chlorophyll a (Chla) variations and associated relationships with environmental stressors along the temporal hydrological connectivity gradient were investigated using a 11-year dataset in a large floodplain lake (Poyang Lake). A geostatistical method was firstly used to calculate the hydrological connectivity curves for each sampling campaign that was further classified by K-means technique. Linear mixed effect (LME) models were developed through the inclusion of the site as a random effect to identify the limiting factors of Chla variations. The results identified three clear hydrological connectivity variation patterns with remarkable connecting water area changes in Poyang Lake. Furthermore, hydrological connectivity changes exerted a great influence on environmental variables in Poyang Lake, with a decrease in nutrient concentrations as the hydrological connectivity enhanced. The Chla exhibited contrast variations with nutrient variables along the temporal hydrological connectivity gradient and generally depended on WT, DO, EC and TP, for the entire study period. Nevertheless, the relative roles of nutrient and non-nutrient variables in phytoplankton growth varied with different degrees of hydrological connectivity as confirmed by the LME models. In the low hydrological connectivity phase, the Chla dynamics were controlled only by water temperature with sufficient nutrients available. In the high hydrological connectivity phase, the synergistic influences of both nutrient and physical variables jointly limited the Chla dynamics. In addition, a significant increasing trend was observed for Chla variations from 2008 to 2018 in the HHC phase, which could largely be attributed to the elevated nutrient concentrations. This study confirmed the strong influences of hydrological connectivity on the nutrient and non-nutrient limitation of phytoplankton growth in floodplain lakes. The present study could provide new insights on the driving mechanisms underlying phytoplankton growth in floodplain lakes.

Novel predictors related to hysteresis and baseflow improve predictions of watershed nutrient loads: An example from Ontario's lower Great Lakes basin

Eutrophication has re-emerged in the lower Great Lakes basin resulting in critical water quality issues. Models that accurately predict nutrient loading from streams are needed to inform appropriate nutrient management decisions. Generalized additive models (GAMs) that use surrogate data from sensors to predict nutrient loads offer an alternative to commonly applied linear regression and may better handle relationship non-linearities and skewed water quality data. Five years (2015-2020) of water quantity and quality data from 11 agricultural watersheds in southern Ontario were used to develop GAMs to predict total phosphorus (TP) and nitrate (NO₃^-) loads. This study aimed to 1) use GAMs to predict nutrient loads using both common and novel predictors and 2) quantify and examine the variability in seasonal and annual nutrient loads. Along with routine surrogate model predictors (i.e., flow, turbidity, and seasonality), the addition of the baseflow proportion and the hydrograph position of flow observations improved model performance. Conversely, including the antecedent precipitation index minimally affected model performance, regardless of constituent. Seasonal and annual patterns in TP and NO₃^- load predictions mirrored that of the hydrologic regime. This study showed that parsimonious GAMs featuring novel model predictors can be used to predict nutrient loads while accounting for the partitioning of surface and subsurface flow paths and hysteresis between streamflow and water quality parameters that are frequently observed in a wide range of environments.

The Spatiotemporal Characteristics of Water Quality and Main Controlling Factors of Algal Blooms in Tai Lake, China

Taking Tai Lake in China as the research area, a 3D water environment mathematical model was built. Combined with the LHS and Morris uncertainty and sensitivity analysis methods, the uncertainty and sensitivity analysis of total phosphorus (TP), total nitrogen (TN), dissolved oxygen (DO), and chlorophyll a (Chl-a) were carried out. The main conclusions are: (1) The performance assessment of the 3D water environment mathematical model is good (R2 and NSE > 0.8) and is suitable for water quality research in large shallow lakes. (2) The time uncertainty study proves that the variation range of Chl-a is much larger than that of the other three water quality parameters and is more severe in summer and autumn. (3) The spatial uncertainty study proves that Chl-a is mainly present in the northwest lake area (heavily polluted area) and the other three water quality indicators are mainly present in the center. (4) The sensitivity results show that the main controlling factors of DO are ters (0.15) and kmsc (0.12); those of TN and TP are tetn (0.58) and tetp (0.24); and those of Chl-a are its own growth rate (0.14), optimal growth temperature (0.12), death rate (0.12), optimal growth light (0.11), and TP uptake rate (0.11). Thus, TP control is still the key treatment method for algal blooms that can be implemented by the Chinese government.

Multi-class secondary metabolites in cyanobacterial blooms from a tropical water body: Distribution patterns and real-time prediction

Cyanobacterial blooms that produce toxins occur in freshwaters worldwide and yet, the occurrence and distribution patterns of many cyanobacterial secondary metabolites particularly in tropical regions are still not fully understood. Moreover, predictive models for these metabolites by using easily accessible water quality indicators are rarely discussed. In this study, we investigated the co-occurrence and spatiotemporal trends of 18 well-known and less-studied cyanobacterial metabolites (including [D-Asp3] microcystin-LR (DM-LR), [D-Asp3] microcystin-RR (DM-RR), microcystin-HilR (MC-HilR), microcystin-HtyR (MC-HtyR), microcystin-LA (MC-LA), microcystin-LF (MC-LF), microcystin-LR (MC-LR), microcystin-LW (MC-LW), microcystin-LY (MC-LY), microcystin-RR (MC-RR) and microcystin-WR (MC-WR), Anatoxin-a (ATX-a), homoanatoxin-a (HATX-a), cylindrospermospin (CYN), nodularin (NOD), anabaenopeptin A (AptA) and anabaenopeptin B (AptB)) in a tropical freshwater lake often plagued with blooms. Random forest (RF) models were developed to predict MCs and CYN and assess the relative importance of 22 potential predictors that determined their concentrations. The results showed that 11 MCs, CYN, ATX-a, HATX-a, AptA and AptB were found at least once in the studied water body, with MC-RR and CYN being the most frequently occurring, intracellularly and extracellularly. AptA and AptB were detected for the first time in tropical freshwaters at low concentrations. The metabolite profiles were highly variable at both temporal and spatial scales, in line with spatially different phytoplankton assemblages. Notably, MCs decreased with the increase of CYN, possibly revealing interspecific competition of cyanobacteria. The rapid RF prediction models for MCs and CYN were successfully developed using 4 identified drivers (i.e., chlorophyll-a, total carbon, rainfall and ammonium for MCs prediction; and chloride, total carbon, rainfall and nitrate for CYN prediction). The established models can help to better understand the potential relationships between cyanotoxins and environmental variables as well as provide useful information for making policy decisions.

Identifying key drivers of harmful algal blooms in a tributary of the Three Gorges Reservoir between different seasons: Causality based on data-driven methods

Intense harmful algal blooms (HABs) can occur in the backwaters of tributaries supplying large-scale reservoirs. Due to the characteristics of process-based models and difficulties in modelling complex nonlinear processes, traditional models have difficulties disentangling the driving factors of HABs. In this study, we used data-driven methods (i.e., correlation analysis and machine-learning models) to identify the most important drivers of HABs in the Xiangxi River, a tributary of the Three Gorges Reservoir, China (2017-2018), for the dry season (from October to mid-April) and wet season (from April to September). We utilized the maximal information coefficient (MIC) combined with a time lag strategy and prior knowledge to quantitatively identify the driving variables of HABs. An extra trees regression (ETR) model was developed to assess the relative importance of causal variables driving algal blooms for the different periods. The results showed that water temperature was the most important driver for the duration of the study, followed by total nitrogen. Nitrogen had a stronger effect on algal blooms than phosphorus during both the wet and dry seasons. HABs were mainly affected by ammonia nitrogen in the wet season and by other forms of nitrogen in the dry season. In contrast, rather than the water temperature and nutrients, the operation of the Three Gorges Dam (difference between inflow and outflow discharge rate) was the most significant factor for algal blooms during the dry season, but its influence sharply declined during the wet season. This study showed that the key drivers of HABs can differ between seasons and suggests that HAB management should take seasonality into account.

Water clarity response to climate warming and wetting of the Inner Mongolia-Xinjiang Plateau: A remote sensing approach

Water clarity (generally quantified as the Secchi disk depth: SDD) is a key variable for assessing environmental changes in lakes. Using remote sensing we calculated and elucidated the SDD dynamics in lakes in the Inner Mongolia-Xinjiang Lake Zone (IMXL) from 1986 to 2018 in response to variations in temperature, rainfall, lake area, normalized difference vegetation index (NDVI) and Palmer's drought severity index (PDSI). The results showed that the lakes with high SDD values are primarily located in the Xinjiang region at longitudes of 75°-93° E. In contrast, the lakes in Inner Mongolia at longitudes of 93°-118° E generally have low SDD values. In total, 205 lakes show significant increasing SDD trends (P < 0.05), with a mean rate of 0.15 m per decade. In contrast, 75 lakes, most of which are located in Inner Mongolia, exhibited significant decreasing trends with a mean rate of 0.08 m per decade (P < 0.05). Pooled together, an overall increase is found with a mean rate of 0.14 m per decade. Multiple linear regression reveals that among the five variables selected to explain the variations in SDD, lake area accounts for the highest proportion of variance (25%), while temperature and rainfall account for 12% and 10%, respectively. In addition, rainfall accounts for 52% of the variation in humidity, 8% of the variation in lake area and 7% of the variation in NDVI. Temperature accounts for 27% of the variation in NDVI, 39% of the variation in lake area and 22% of the variation in PDSI. Warming and wetting conditions in IMXL thus promote the growth of vegetation and cause melting of glaciers and expansion of lake area, which eventually leads to improved water quality in the lakes in terms of higher SDD. In contrast, lakes facing more severe drought conditions, became more turbid.

Impacts of storm events on chlorophyll-a variations and controlling factors for algal bloom in a river receiving reclaimed water

Harmful algal bloom is prevalent in the reclaimed-water-source (RWS) river caused by the excessive nutrient's inputs. Rainfall water may be the sole nutrient-diluted water source for the RWS river. However, the effects of storm events on the algal bloom in the RWS river are poorly understood. This study presents chlorophyll-a (Chl-a) variations before, during, and after the initial storm events (Pre-storm, In-storm, and Post-storm) at four representative sites with distinct hydraulic conditions in a dam-regulated RWS river system, Beijing. The response of Chl-a to the initial storm events mostly depends on the ecosystem status that caused by the river hydraulic properties. The upstream is more river-like and downstream is more lake-like. In the river-like system, elevated water temperature (WT, increased by 2 %) could support the dominating algae (diatom) growth (Chl-a increased by 130 %) from Pre-storm to In-storm period. In the lake-like system, the dominant algae (blue algae) declined (Chl-a sharply decreased by 96%-99 %) due to the lower WT (decreased by 3%-7%) and increased flow velocities from Pre-storm to In-storm period. During the Post-storm period, the dominant algae break out (Chl-a surged by 20%-319 %) in the lake-like system caused by the recovered WT (increased by 3%-6%) and flow velocity. The occurrence of algal bloom can be predicted by the Random Forest (RF) model based on water quality parameters such as total nitrogen (TN). The thresholds of algal bloom for the Pre-storm, In-storm, and Post-storm periods were identified as 30 μg/L, 10 μg/L, and 10 μg/L, respectively. The two driven factors were WT and nitrate (NO₃-N) for the Pre-storm period and were WT and TN for the In- & Post-storm periods. A higher risk of algal bloom is highlighted during the initial storm events in the RWS river. We propose recommendations for improving water quality in the RWS river systems under the climatic change.

Evaluation of Machine Learning Predictions of a Highly Resolved Time Series of Chlorophyll-a Concentration

Pelagic chlorophyll-a concentrations are key for evaluation of the environmental status and productivity of marine systems, and data can be provided by in situ measurements, remote sensing and modelling. However, modelling chlorophyll-a is not trivial due to its nonlinear dynamics and complexity. In this study, chlorophyll-a concentrations for the Helgoland Roads time series were modeled using a number of measured water and environmental parameters. We chose three common machine learning algorithms from the literature: the support vector machine regressor, neural networks multi-layer perceptron regressor and random forest regressor. Results showed that the support vector machine regressor slightly outperformed other models. The evaluation with a test dataset and verification with an independent validation dataset for chlorophyll-a concentrations showed a good generalization capacity, evaluated by the root mean squared errors of less than 1 µg L−1. Feature selection and engineering are important and improved the models significantly, as measured in performance, improving the adjusted R2 by a minimum of 48%. We tested SARIMA in comparison and found that the univariate nature of SARIMA does not allow for better results than the machine learning models. Additionally, the computer processing time needed was much higher (prohibitive) for SARIMA.

Delineating the relative contribution of climate related variables to chlorophyll-a and phytoplankton biomass in lakes using the ERA5-Land climate reanalysis data

Understanding the climatic drivers of eutrophication is critical for lake management under the prism of the global change. Yet the complex interplay between climatic variables and lake processes makes prediction of phytoplankton biomass a rather difficult task. Quantifying the relative influence of climate-related variables on the regulation of phytoplankton biomass requires modelling approaches that use extensive field measurements paired with accurate meteorological observations. In this study we used climate and lake related variables obtained from the ERA5-Land reanalysis dataset combined with a large dataset of in-situ measurements of chlorophyll-a and phytoplankton biomass from 50 water bodies to develop models of phytoplankton related responses as functions of the climate reanalysis data. We used chlorophyll-a and phytoplankton biomass as response metrics of phytoplankton growth and we employed two different modelling techniques, boosted regression trees (BRT) and generalized additive models for location scale and shape (GAMLSS). According to our results, the fitted models had a relatively high explanatory power and predictive performance. Boosted regression trees had a high pseudo R² with the type of the lake, the total layer temperature, and the mix-layer depth being the three predictors with the higher relative influence. The best GAMLSS model retained mix-layer depth, mix-layer temperature, total layer temperature, total runoff and 10-m wind speed as significant predictors (p<0.001). Regarding the phytoplankton biomass both modelling approaches had less explanatory power than those for chlorophyll-a. Concerning the predictive performance of the models both the BRT and GAMLSS models for chlorophyll-a outperformed those for phytoplankton biomass. Overall, we consider these findings promising for future limnological studies as they bring forth new perspectives in modelling ecosystem responses to a wide range of climate and lake variables. As a concluding remark, climate reanalysis can be an extremely useful asset for lake research and management.

Stochastic simulation of phytoplankton biomass using eighteen years of daily data - predictability of phytoplankton growth in a large, shallow lake

During the past decades, on-line monitoring of freshwater lakes has developed rapidly. To use high frequency time-series in lake management, novel models are needed that are simple and provide insight into the complexity of phytoplankton dynamics. Chlorophyll a (Chl), a proxy for phytoplankton biomass and environmental drivers were monitored on-line in large, shallow Lake Balaton during the vegetation periods between 2001 and 2018. Growth and non-growth (G and non-G) states of algae were deduced from daily change in Chl. Random forests (RF) were used to find stochastic response rules of phytoplankton to growth-supporting environmental habitat templates. The stochastic G/non-G state was translated into long-term daily biomass dynamics by a deterministic biomass model to assess uncertainty and to distinguish between inevitable and unpredictable blooms. A biomass peak was qualified as inevitable or unpredictable if the lower 95% confidence limit of simulations exceeded or remained at the baseline Chl level, respectively. Compared to a stochastic null model based on monthly Markovian transition probabilities, RF-based models captured wax and wane of biomass realistically. Timing of peaks could be better simulated than their magnitude, likely because habitat templates were primarily determined by light whereas peak sizes might depend on unmeasured processes, such as phosphorus availability. In general, algal growth was favored by wind-induced sediment resuspension that decreased light availability but simultaneously enhanced the P supply. Seasonal temperature and an integral of departures from the "normal" seasonal temperature over 2 to 3 generations were important drivers of phytoplankton growth, whereas short-term (diel and day to day) changes in water temperature appeared to be irrelevant. Four types of years could be distinguished during the study period with respect to algal growth conditions. The present modeling approach can reasonably be used even in highly variable aquatic environments when 3 to 4 years of daily data are available.

Assessing Phytoplankton Bloom Phenology in Upwelling-Influenced Regions Using Ocean Color Remote Sensing

Phytoplankton bloom phenology studies are fundamental for the understanding of marine ecosystems. Mismatches between fish spawning and plankton peak biomass will become more frequent with climate change, highlighting the need for thorough phenology studies in coastal areas. This study was the first to assess phytoplankton bloom phenology in the Western Iberian Coast (WIC), a complex coastal region in SW Europe, using a multisensor long-term ocean color remote sensing dataset with daily resolution. Using surface chlorophyll a (chl-a) and biogeophysical datasets, five phenoregions (i.e., areas with coherent phenology patterns) were defined. Oceanic phytoplankton communities were seen to form long, low-biomass spring blooms, mainly influenced by atmospheric phenomena and water column conditions. Blooms in northern waters are more akin to the classical spring bloom, while blooms in southern waters typically initiate in late autumn and terminate in late spring. Coastal phytoplankton are characterized by short, high-biomass, highly heterogeneous blooms, as nutrients, sea surface height, and horizontal water transport are essential in shaping phenology. Wind-driven upwelling and riverine input were major factors influencing bloom phenology in the coastal areas. This work is expected to contribute to the management of the WIC and other upwelling systems, particularly under the threat of climate change.

Effects of basin nutrient discharge variations coupled with climate change on water quality in Lake Erhai, China

In Lake Erhai, water quality was affected by the basin nutrient discharge and climate change. To analyze the relationships between the water quality (total nitrogen [TN], total phosphorus [TP], chemical oxygen demand [CODmn], ammonia [NH₄], and trophic level index [TLI]) and basin nutrient discharge (TNd, TPd, and CODd) combined with climate changes (air temperature [AT], precipitation [pre], wind speed [wind], and sunshine hours [SHs]), the generalized additive model (GAM) was employed to explore the nonlinear relationships with their interactions using data sets ranging from 1999 to 2012. Our findings revealed that the water quality in Lake Erhai deteriorated in the early twentieth century, and the basin discharge and AT appeared significant (p < 0.05) rising trends in a long time, while the precipitation decreased significantly (p < 0.05) in the study period. Single-factor GAM results indicated that the basin nutrient discharge was the main explanatory factor for the variations of TN and TP in lake, while precipitation was the main driver for CODmn and NH₄. Besides, the water quality displayed nonlinear responses to the basin discharge, but all of the water quality variables went up as the emission levels increased in the lower range. The results showed that the water quality deteriorated in the lower rainfall, and TN rose as the AT increases, while TP was elevated accompanied by the ascending SHs there. The GAM interaction results suggested that the increase of AT and TPd had a promoting effect on TP in Lake Erhai. Stricter nutrient management measures should be implemented when the impacts of climate change are taken into account.

Environmental assessment of physical-chemical features of Lake Nasser, Egypt

Lake Nasser is one of the largest man-made lakes on earth. It has a vital importance to Egypt for several decades because of the safe water supply of the country. Therefore, the water quality of the Lake Nasser must be profoundly investigated, and physico-chemical parameter changes of the water of the Lake Nasser should be continuously monitored and assessed. This work describes the present state of the physico-chemical (nitrate-nitrogen, nitrite-nitrogen, orthophosphate, total phosphate content, dissolved oxygen content, chemical oxygen demand, and biological oxygen demand) water parameters of Lake Nasser in Egypt at nine measurement sites along the Lake Nasser. The algorithm was devised at the University of Pannonia, Hungary, for the evaluation of the water quality. The aquatic environmental indices determined alongside the Lake Nasser fall into the category of "good" water quality at seven sampling sites and exhibited "excellent" water quality at two sampling sites according to Egyptian Governmental Decree No. 92/2013. In light of the tremendous demand for safe and healthy water supply in Egypt and international requirements, the water quality assessment is a very important tool for providing reliable information on the water quality. The protocol for water quality assessment could significantly contribute to the provision of high-quality water supply in Egypt. In conclusion, it can be stated that the parameters under investigation in different regions of Lake Nasser fall within the permissible ranges and the water of the Lake has good quality for drinking, irrigation, and fish cultures according to Egyptian standards; however, according to European specifications, there are steps to be accomplished for future water quality improvement.

An ensemble modeling framework to study the effects of climate change on the trophic state of shallow reservoirs

Our understanding of the potential impact of climatic change on catchment hydrology and aquatic system dynamics has been advanced over the past decade, but there are still considerable knowledge gaps with respect to its effects on water quality vis-à-vis the increasing demands for drinking water. In this study, we developed an integrated hydrological-water quality (SWAT-YRWQM) model to elucidate the effects of a changing climate on the trophic state of the shallow Yuqiao Reservoir. Using a two-step downscaling process, we reproduced the prevailing meteorological conditions, as well as the streamflows in three major tributaries of the study area. A sensitivity analysis exercise showed that the nature of the calibration dataset used, namely the range of flows (i.e., dry versus wet years) included, can profoundly influence the predictive power of our modeling framework. Our climatic scenarios projected a minor change of the streamflow rates, but a variant degree of increase of the riverine total phosphorus (TP) concentrations and associated loading rates into the reservoir. Consequently, a significant rise of in-lake TP concentrations is projected for the near (2016-2030) and distant (2031-2050) future compared to the reference (2006-2015) conditions. Interestingly, the ambient TP levels appear to be lower in the distant relative to the near future, owing to changes in the magnitude and relative contribution of both external and internal nutrient loading sources. Our analysis also highlights the importance of reservoir operation practices to regulate water levels as a means for mitigating the climate change impact on the trophic status of the Yuqiao Reservoir, given that the diversion of low-nutrient water from the upstream basin can significantly reduce (30-40%) the TP concentrations. Our findings are highly relevant to the on-going debate about the potential implications of climate change for water availability, highlighting the importance of adaptation strategies to optimize the water resources management.