Many studies have attempted to predict chlorophyll-a concentrations using multiple regression models and validating them with a hold-out technique. In this study commonly used machine learning models, such as Support Vector Regression, Bagging, Random Forest, Extreme Gradient Boosting (XGBoost), Recurrent Neural Network (RNN), and Long–Short-Term Memory (LSTM), are used to build a new model to predict chlorophyll-a concentrations in the Nakdong River, Korea. We employed 1–step ahead recursive prediction to reflect the characteristics of the time series data. In order to increase the prediction accuracy, the model construction was based on forward variable selection. The fitted models were validated by means of cumulative learning and rolling window learning, as opposed to the hold–out technique. The best results were obtained when the chlorophyll-a concentration was predicted by combining the RNN model with the rolling window learning method. The results suggest that the selection of explanatory variables and 1–step ahead recursive prediction in the machine learning model are important processes for improving its prediction performance.
[1] We investigated the sources, fluxes, and spatial and temporal distribution of organic carbon in two large reservoirs in Korea having different trophic and hydrologic features. In Lake Paldang (river-type reservoir) most of the organic carbon comes from allochthonous sources (88%), while in Lake Chungju (lake-type reservoir) the allochthonous load contributed 52% (48% autochthonous). Strong correlations were found between the amount and contribution of allochthonous and autochthonous organic carbon and hydrologic parameters in Lake Paldang but not in Lake Chungju. The spatiotemporal fluctuation patterns of C/N ratios of particulate organic matter, chlorophyll a/particulate organic carbon values, and specific ultraviolet absorption of dissolved organic carbon were different, indicating differences in state and source of organic carbon in the two reservoirs. Our results indicate that meteorologic and hydrologic controls directly determine the state of lacustrine organic carbon in Lake Paldang, whereas autochthonous production and in situ transformation may play important roles in determining the state of lacustrine organic carbon in Lake Chungju.
Despite the implementation
of intensive phosphorus reduction measures,
periodic outbreaks of cyanobacterial blooms in large rivers remain
a problem in Korea, raising the need for more effective solutions
to reduce their occurrence. This study sought to evaluate whether
phosphorus or nitrogen limitation is an effective approach to control
cyanobacterial (Microcystis) blooms in river conditions
that favor this non-nitrogen-fixing genus. These conditions include
nutrient enrichment, high water temperature, and thermal stratification
during summer. Mesocosm bioassays were conducted to investigate the
limiting factors for cyanobacterial blooms in a river reach where
severe Microcystis blooms occur annually. We evaluated
the effect of five different nitrogen (3, 6, 9, 12, and 15 mg/L) and
phosphorus (0.01, 0.02, 0.05, 0.1, and 0.2 mg/L) concentrations on
algae growth. The results indicate that nitrogen treatments stimulated
cyanobacteria (mostly Microcystis aeruginosa) more
than phosphorus. Interestingly, phosphorus additions did not stimulate
cyanobacteria, although it did stimulate Chlorophyceae and Bacillariophyceae.
We conclude that phosphorus reduction might have suppressed the growth
of Chlorophyceae and Bacillariophyceae more than that of cyanobacteria;
therefore, nitrogen or at least both nitrogen and phosphorus control
appears more effective than phosphorus reductions alone for reducing
cyanobacteria in river conditions that are favorable for non-nitrogen-fixing
genera.
Blooms of harmful cyanobacteria Microcystis aeruginosa lead to an adverse effect on freshwater ecosystems, and thus extensive studies on the control of this cyanobacteria’s blooms have been conducted. Throughout this study, we have found that the two bacteria Aeromonas bestiarum HYD0802-MK36 and Pseudomonas syringae KACC10292T are capable of killing M. aeruginosa. Interestingly, these two bacteria showed different algicidal modes. Based on an algicidal range test using 15 algal species (target and non-target species), HYD0802-MK36 specifically attacked only target cyanobacteria M. aeruginosa, whereas the algicidal activity of KACC10292T appeared in a relatively broad algicidal range. HYD0802-MK36, as a direct attacker, killed M. aeruginosa cells when direct cell (bacterium)-to-cell (cyanobacteria) contact happens. KACC10292T, as an indirect attacker, released algicidal substance which is located in cytoplasm. Interestingly, algicidal activity of KACC10292T was enhanced according to co-cultivation with the host cyanobacteria, suggesting that quantity of algicidal substance released from this bacterium might be increased via interaction with the host cyanobacteria.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.