Elucidation and short-term forecasting of microcystin concentrations in Lake Suwa (Japan) by means of artificial neural networks and evolutionary algorithms

doi:10.1016/j.watres.2007.02.001

Water Research

Volume 41, Issue 10, May 2007, Pages 2247-2255

https://doi.org/10.1016/j.watres.2007.02.001 Get rights and content

Abstract

Non-supervised artificial neural networks (ANN) and hybrid evolutionary algorithms (EA) were applied to analyse and model 12 years of limnological time-series data of the shallow hypertrophic Lake Suwa in Japan. The results have improved understanding of relationships between changing microcystin concentrations, Microcystis species abundances and annual rainfall intensity. The data analysis by non-supervised ANN revealed that total Microcystis abundance and extra-cellular microcystin concentrations in typical dry years are much higher than those in typical wet years. It also showed that high microcystin concentrations in dry years coincided with the dominance of the toxic Microcystis viridis whilst in typical wet years non-toxic Microcystis ichthyoblabe were dominant. Hybrid EA were used to discover rule sets to explain and forecast the occurrence of high microcystin concentrations in relation to water quality and climate conditions. The results facilitated early warning by 3-days-ahead forecasting of microcystin concentrations based on limnological and meteorological input data, achieving an r²=0.74 for testing.

Introduction

The shallow eutrophic Lake Suwa is being studied for decades with regards to its distinctive patterns of high abundances of Microcystis species and high concentrations of microcystin during summer (e.g. Harada et al., 2001; Park et al., 1993, Park et al., 1998). In situ measurements of limnological variables of Lake Suwa from 1992 to 2003 provided precious information about complex relationships between water quality, meteorology and Microcystis population dynamics.

Non-supervised artificial neural networks (ANN) have been demonstrated to be useful tools for ordination, clustering and visualisation of complex data such as of water treatment plants, stream and lake habitats (e.g. Hong et al., 2003; Park et al., 2003; Recknagel et al., 2006). Supervised ANN and evolutionary algorithms (EA) have successfully been applied to facilitate early warning of sudden outbreaks of toxic blue-green algae in freshwater lakes and rivers by time-series modelling (e.g. Recknagel et al., 1997, Recknagel et al., 2002; Cao et al., 2006; Jeong et al., 2006). However there is no example yet known for ordination, clustering and forecasting microcystin concentrations in natural water bodies using water quality time-series data.

The present study utilised non-supervised ANN and hybrid EA for unravelling complex ecological relationships in the database of Lake Suwa, and forecasting of microcystin concentrations by means of water quality and meteorological data. The study aimed at following research questions: (i) can non-supervised ANN determine seasonal relationships amongst water quality data, Microcystis species and microcystin concentrations with regards to rainfall patterns, (ii) can hybrid EA forecast 3-days ahead the timing and magnitudes of extra-cellular microcystin concentrations, and (iii) can hybrid EA discover rule sets to describe conditions for microcystin occurrence in Lake Suwa.

Section snippets

Study site and data

The study site of this research was Lake Suwa located in central Japan. Lake Suwa is a shallow, hypertrophic lake with occasional weak thermal stratification in summer between April and October and frequent ice cover in winter between December and February. Lake Suwa is mainly used for recreational activities, but a small proportion of lake water is also used for aquaculture, irrigation and industrial purposes (Park et al., 1998).

Microcystis blooms in Lake Suwa have been reported every summer

Ordination and clustering of water quality data by non-supervised ANN

The ordination and clustering of water quality data of Lake Suwa by means of non-supervised ANN has been conducted according to Fig. 1, Fig. 2 distinguishing between typical dry years (1992 and 1994) and wet years (1999 and 2000).

Results in Fig. 4 indicate distinctive differences both in magnitudes as well as seasonal occurrences of Microcystis abundances between dry and wet years. Whilst maximum cell numbers peak in the dry years at 515,000 cells/mL in autumn (September and October), they peak

Ordination and clustering

Non-supervised ANN have unravelled complex relationships between Microcystis species and microcystin concentrations for different seasons and rainfall intensities. The big differences in the abundance of Microcystis and therefore microcystin concentration between typical dry and typical wet years seem to reflect the impact of the Asian monsoon in summer. As reported by Ahn and Jones (2000) extreme cyanobacterial blooms are unlikely to develop in Korean freshwaters during summers with intense

Conclusions

Non-supervised ANN successfully revealed the seasonal succession of Microcystis species and microcystin dynamics in relation to different rainfall and environmental conditions of Lake Suwa. Hybrid EA assembled rule sets that proved to be valid for both prediction and explanation of microcystin dynamics. Outcomes of this research can facilitate early warning of high microcystin concentrations in Lake Suwa.

References (22)

Y.-S.T. Hong et al.
Analysis of a municipal wastewater treatment plant using a neural network-based pattern analysis
Water Res.
(2003)
Y.-S. Park et al.
Patterning and predicting aquatic macroinvertebrate diversities using artificial neural network
Water Res.
(2003)
F. Recknagel et al.
Artificial neural network approach for modelling and prediction of algal blooms
Ecol. Modelling
(1997)
F. Recknagel et al.
Unravelling and forecasting algal population dynamics in two lakes different in morphometry and eutrophication by neural and evolutionary computation
Ecol. Informatics
(2006)
D. Whitall et al.
Importance of atmospherically deposited nitrogen to the annual nitrogen budget of the Neuse River estuary, North Carolina
Environ. Int.
(2003)
K.-G. Ahn et al.
Factors regulating bluegreen dominance in a reservoir directly influenced by the Asian monsoon
Hydrobiologia
(2000)
C.-Y. Ahn et al.
Rainfall, phycocyanin, and N:P ratios related to cyanobacterial blooms in a Korean large reservoir
Hydrobiologia
(2002)
H. Cao et al.
Hybrid evolutionary algorithm for rule set discovery in time-series data to forecast and explain algal population dynamics in two lakes different in morphometry and eutrophication
K. Harada et al.
Co-production of microcystins and aeruginopeptins by natural cyanobacterial bloom
Environ. Toxicol.
(2001)
J.H. Holland
Adaptation in Natural and Artificial Systems
(1975)

K.-S. Jeong et al.

Prediction and elucidation of population dynamics of the blue-green algae Microcystis aeruginosa and the diatom Stephanodiscus hantzschii in the Nakdong river-reservoir system (South Korea) by a recurrent artificial neural network

Cited by (50)

Decision tree ensemble with Bayesian optimization to predict the spatial dynamics of chlorophyll-a concentration: A case study in Bay of Bengal
2024, Marine Pollution Bulletin
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.
Multi-class secondary metabolites in cyanobacterial blooms from a tropical water body: Distribution patterns and real-time prediction
2022, Water Research
Citation Excerpt :
The potential impact of rainfall on toxic cyanobacterial blooms is contradictory, complex and strongly depends on site-specific dynamics, cyanobacterial species composition and strain succession (Reichwaldt and Ghadouani, 2012). The frequent occurrence of heavy rainfall events has been reported to cause either a disruption of the bloom, or a decrease in cyanobacterial mass due to de-stratification, increased turbidity, high turbulence, low irradiance, dilution and flushing, consequently resulting in a decrease in cyanotoxin production (Chan et al., 2007; Reichwaldt and Ghadouani, 2012). In contrast, prolonged rainfall and intense rainfall events may favor cyanobacteria via increased nutrient input into the inland waterbodies (Wood et al., 2017).
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.
Cyanobacterial blooms in China: diversity, distribution, and cyanotoxins
2020, Harmful Algae
Cyanobacterial blooms, which refer to the massive growth of harmful cyanobacteria, have altered the global freshwater ecosystems during the past decades. China has the largest population in the world, and it is suffering from the harmful effect of water eutrophication and cyanobacterial blooms along with rapid development of the economy and society. Research on cyanobacterial blooms and cyanotoxins in China have been overwhelmingly enhanced and emphasized during the past decades. In the present review, the research on cyanobacterial blooms in China is generally introduced, including the history of cyanobacterial bloom studies, the diversity of the bloom-forming cyanobacteria species (BFCS), and cyanotoxin studies in China. Most studies have focused on Microcystis, its blooms, and microcystins. Newly emerging blooms with the dominance of non-Microcystis BFCS have been gradually expanding to wide regions in China. Understanding the basic features of these non-Microcystis BFCS and their blooms, including their diversity, occurrence, physio-ecology, and harmful metabolites, will provide direction on future studies of cyanobacterial blooms in China.
Spatial-temporal process simulation and prediction of chlorophyll-a concentration in Dianchi Lake based on wavelet analysis and long-short term memory network
2020, Journal of Hydrology
With the rapid development of urbanization, the water pollution in Dianchi Lake presents the trend of combining urban and agricultural non-point source pollution, and it is more difficult to control and improve the water environment. The simulation and prediction of water quality state change is an important theoretical basis for water resources management. The data set we selected was 15 water quality parameters of 10 water quality observation sites in Dianchi Lake from 2005 to 2012. Wavelet Domain Threshold Denoising (WDTD), Wavelet Mean Fusion (WMF) and Long-Short Term Memory (LSTM) were combined to a WDTD-LSTM-WMF long-term prediction model that WMF was proposed based on WDTD in this paper. The model and geospatial analysis were used to simulate the historical change process of chlorophyll-a concentration (Chla) in Dianchi Lake and predicted the future trend of Chla. The results showed that the model has a good prediction performance of low error and high generalization (RMSE = 18.40, MAE = 13.56, R² = 0.63). The spatial visualization analysis showed that the region with Chla higher than 100 μg/L from 2005 to 2020 had a tendency to spread from north to west and then to southwest. This is related to the urbanization development and climate change in Kunming.
Editorial: Analysis and synthesis of ecological data by machine learning
2019, Ecological Informatics
Assessing the effect of abiotic variables and zooplankton on picocyanobacterial dominance in two tropical mesotrophic reservoirs by means of evolutionary computation
2019, Water Research
Evolutionary computation has been applied to predict the occurrence of massive cyanobacteria proliferations; in the present study, this tool was further used to explore the factors responsible for maintaining picocyanobacterial dominance. Aiming to increase the understanding of factors that promote dominance of picocyanobacteria in tropical reservoirs, we chose two reservoirs used for water supplies located in different regions of Brazil and subjected to climate changes such as warmer winters that intensify water column stratification and prolonged dry seasons that cause water level decreases. This study focused on the diagnosis of the relationships among picocyanobacteria (1–2 μm), zooplankton and environmental variables using evolutionary computation. The integrated data analysis performed here was very successful in elucidating the dynamics of picocyanobacterial density variation influenced by both abiotic and biotic factors by the modeling approach. Relative water column stability - RWCS and electrical conductivity were highlighted as the most important environmental drivers for picocyanobacterial peaks. Hybrid Evolutionary Analysis (HEA) models for the two reservoirs indicated for the first time in the literature that rotifers, small-sized cladocerans and copepods (mainly nauplii) can directly or indirectly control picocyanobacteria in tropical mesotrophic reservoirs, depending on RWCS conditions and electrical conductivity. However, this control is modulated by pH, water transparency and water temperature thresholds.

View all citing articles on Scopus

View full text

Elucidation and short-term forecasting of microcystin concentrations in Lake Suwa (Japan) by means of artificial neural networks and evolutionary algorithms

Abstract

Introduction

Section snippets

Study site and data

Ordination and clustering of water quality data by non-supervised ANN

Ordination and clustering

Conclusions

Water Res.

Water Res.

Ecol. Modelling

Ecol. Informatics

Environ. Int.

Factors regulating bluegreen dominance in a reservoir directly influenced by the Asian monsoon

Hydrobiologia

Rainfall, phycocyanin, and N:P ratios related to cyanobacterial blooms in a Korean large reservoir

Hydrobiologia

Hybrid evolutionary algorithm for rule set discovery in time-series data to forecast and explain algal population dynamics in two lakes different in morphometry and eutrophication

Co-production of microcystins and aeruginopeptins by natural cyanobacterial bloom

Environ. Toxicol.

Adaptation in Natural and Artificial Systems

Prediction and elucidation of population dynamics of the blue-green algae Microcystis aeruginosa and the diatom Stephanodiscus hantzschii in the Nakdong river-reservoir system (South Korea) by a recurrent artificial neural network