Masayoshi Harada

Assessment of spatial habitat heterogeneity by coupling data-driven habitat suitability models with a 2D hydrodynamic model in small-scale streams

Abstract Habitat assessment considering habitat quality and quantity is a key approach in conservation and restoration works for biodiversity and ecosystems. In this regard, application of hydrodynamic model for modeling instream habitat conditions and machine learning (ML) methods for modeling habitat suitability of a target species can contribute to better modeling practices in ecohydraulics. Despite the importance of small streams for aquatic ecosystems, previous studies in ecohydraulics have been conducted mainly in medium to large rivers, often disregarding small-scale streams such as agricultural canals. The aim of this study is to demonstrate the applicability of a coupled use of ML and a two-dimensional (2D) hydrodynamic model for assessing spatial habitat heterogeneity in small-scale agricultural canals in Japan. Using abundance data of Japanese medaka ( Oryzias latipes ), four ML methods, namely artificial neural networks (ANNs), classification and regression trees (CARTs), random forests (RF) and support vector machines (SVMs), were applied to develop habitat suitability models considering water depth and flow velocity. A 2D hydrodynamic model was developed based on field surveys in two types of agricultural canals, namely earthen and concrete-lined canals. Information entropy was used for assessing the spatial heterogeneity of instream habitat conditions. As a result, the hydrodynamic models could model instream habitat conditions in a reasonable accuracy. Despite the differences in accuracies in habitat modeling, the four ML methods illustrated similar habitat suitability information for Japanese medaka. The coupled ecohydraulics modeling approach could quantify habitat quality and its spatial heterogeneity, based on which the differences between the earthen and concrete-lined canals were quantitatively assessed. This study demonstrated the applicability of ML-based habitat suitability evaluation and a 2D hydrodynamic model for modeling the spatial distribution of habitat suitability and assessing its spatial heterogeneity. Further study, assessing the spatial heterogeneity in various types of flows including natural/artificial and small/large streams, can contribute to establish quantitative criteria for an ecologically sound habitat and improved ecofriendly construction works in small-scale rivers and streams.

Evaluation of the water-quality dynamics in a eutrophic agricultural pond by using a one-box ecosystem model considering several algal groups

The Okubo pond is an agricultural pond located in the Itoshima area of Fukuoka Prefecture, Japan. It was constructed for irrigating a nearby cultivated area. A monitoring program from August 11 to November 26, 2008, suggested the risk of eutrophication in the water body. The high total nitrogen concentration (1.34 mg/l), high total phosphorus concentration (0.06 mg/l) and extreme oxygen depletion (2 mg/l on 5 November 2008) exceeded the Japanese standards for paddy irrigation water. Furthermore, luxuriant algal blooming in October indicated a hypereutrophic status according to the OECD (1982) criteria. In this study, a one-box ecosystem model was developed to obtain an insight into the seasonal variations in the algal concentrations and chemical components of the Okubo pond with the aims of protecting its aquatic ecosystem and maintaining its water quality. The model was based on a completely mixed system and included 12 water-quality indices: green algae, blue-green algae, diatoms and dinoflagellates, cryptophytes, zooplankton, particulate organic matter, dissolved organic matter, phosphate–phosphorus, ammonia–nitrogen, nitrite–nitrogen, nitrate–nitrogen, and dissolved oxygen. The monitoring data were used to verify the model simulation. Model evaluation suggested good agreement between the predicted and the observed data for the seasonal variations in the algal, nutrient, and dissolved oxygen concentrations. To determine the sensitivity of the model parameters, a sensitivity analysis was conducted leading to the conclusion that the optimal temperature, growth rate, and respiratory rate of the four algal groups, especially the ideal temperature for blue-green and green algae are the most influential factors determining the variability affecting the model inference. From the model simulation, the water temperature and nutrient concentration were considered to be the key limiting factors controlling alternative algal blooms in this period.

Simulation of rainfall runoff and pollutant load for Chikugo River basin in Japan using a GIS-based distributed parameter model

In watershed management, the determination of peak and total runoff due to rainfall and prediction of pollutant load are very important. Measurement of rainfall runoff and pollutant load is always the best approach but is not always possible at the desired time and location. In practice, diffuse pollution has a complex natural dependence on various land-use activities such as agriculture, livestock breeding, and forestry. Estimation of pollutant load is therefore essential for watershed management and water pollution control. In this study, a model of rainfall runoff and pollutant load, which uses a geographical information system (GIS) database, is a convenient and powerful tool for resolving the abovementioned complexities. This technology was applied in order to simulate the runoff discharge and the pollutant load of total nitrogen (TN) and total phosphorus (TP) in the Chikugo River basin of Kyushu Island, Japan. First, a hydrologic modeling system (HEC-HMS) and GIS software extension tool were used for simulations of elevation, drainage line definition, watershed delineation, drainage feature characterization, and geometric network generation. The spatial distributions of land cover, soil classes, rainfall, and evaporation were then analyzed in order to simulate the daily runoff discharge at the Chikugo Barrage from April 2005 to December 2007. An important point in this approach is that a new development for data input processing with HEC-HMS was introduced for optimizing parameters of the model. Next, the water quality indicators TN and TP were examined, and an efficient approach was investigated for estimating monthly pollutant loads directly from unit load and ground-observed hydrological data. Both nonpoint and point sources of pollutants were considered, including different land-cover categories, sewers, factories, and livestock farms. The observed and simulated results for the runoff discharges and pollutant loads were in good agreement and totally consistent, indicating that the proposed model is applicable to simulation of rainfall runoff and pollutant load in the Chikugo River basin. Further, this model will be able to provide managers with a useful tool for optimizing the water surface management of this river basin.

Generation of hydrogen sulfide in the deepest part of a reservoir under anoxic water conditions

The generation of hydrogen sulfide in a reservoir can be characterized by a massive influx of organic matter, manifest thermal stratification, and an extreme anoxic state specifically at the hypolimnion layer in summer. It has been reported that in this deep layer of a thermally stratified water body, where dissolved oxygen is depleted and redox potential falls sharply below the value of zero, the prolonged oxygen-free period and negative redox potential may potentiate the production of toxic hydrogen sulfide. The main aim of this study was to simulate the varying concentration of hydrogen sulfide at the deepest part of the reservoir based on intensive monitoring of water quality parameters. A one-dimensional hydraulic ecosystem model was applied to simulate vertical distribution of dissolved oxygen (DO) continuously in time. Based on the simulated DO, the length of continuous oxygen-free state in the deepest layer was specified and then utilized to estimate the temporal changes of oxidation–reduction potential (ORP) using a polynomial regression model. Finally, a linear regression model for hydrogen sulfide was fitted from the observed ORP, based on which, the temporal changes of hydrogen sulfide concentration were modeled using the simulated ORP. The introduction of several calculations gave good simulated results of hydrogen sulfide concentration, and this approach can be considered useful for predictive and explanatory purposes in further research investigating toxic hydrogen sulfide at the hypolimnion of the reservoir.

Application of a simple genetic algorithm for the calibration of aquatic ecosystem model of an agricultural pond

In this study, we aim to construct and apply a simple genetic algorithm (SGA) to optimize a large number of parameters of an one-box ecosystem model. The ecosystem model was used to simulate the water quality over a 6-month period based on the new observation data in an agricultural pond which was strongly influenced by a green algal bloom. Of the 54 parameters in this model, 10 important parameters were initially selected for the optimization, with one other parameter being subsequently added. The SGA program was used for three purposes, namely (1) to narrow the search space for the 10 parameters, (2) to assess the influence of the additional parameter on the optimization results, and (3) to observe the distribution and convergence of the optimized values for the 10 selected parameters. In the next step, new ranges for these 10 important parameters were assigned and the SGA was applied to all 54 model parameters to seek the optimum value for each parameter. The modeling approach and the results presented here provide valuable and reliable evidences of the optimum parameters for further simulations to clarify the mechanisms of the biochemical processes in the water.

Experimental study on the influence of dissolved organic matter in water and redox state of bottom sediment on water quality dynamics under anaerobic conditions in an organically polluted water body

To understand aquatic environmental deterioration due to the anoxic state in an organically polluted water body, water quality dynamics under anaerobic conditions were examined through beaker-scale water quality monitoring. This study focused on the dynamic properties of NO3–N, NH4–N, PO4–P, and sulfide from the biochemical reactions point of view, and based on anaerobic respiration activities, such as denitrification, iron reduction, and sulfate reduction. The specific aims of this study were to quantitatively estimate the impacts of the oxidative/reductive state of the sediment surface and the high/low concentrations of dissolved organic matter on the dynamic properties of water quality under anaerobic conditions. The beaker-scale water quality monitoring was carried out through continuous measurements of dissolved oxygen and oxidation–reduction potential (ORP), as well as periodic observations of water quality parameters for six cases that are composed from combinations of three experimental conditions: (1) concentration of dissolved organic carbon (DOC); (2) redox state of the sediment; and (3) concentration of NO3–N. As a result, the temporal changes in ORP under anaerobic conditions exhibited a five-step process of decline without these conditions. Also, high DOC concentrations and oxidative states of the bottom mud accelerated the release of PO4–P and sulfide generation through promoting biological iron reduction and sulfate reduction, because these conditions increase respiratory substrates for anaerobic microorganisms, such as iron and sulfate-reducing bacteria. It was concluded that our results would provide important information about the mechanisms of aqueous environmental deterioration due to organic pollution in closed water bodies.

Sensitivity-Based calibration of the soil and water assessment tool for hydrologic cycle simulation in the cong watershed, Vietnam

For better water resources management in quality and quantity, an accurate model is needed for simulating the hydrologic cycle and water quality dynamics at a watershed scale. In this study, a semi-distributed hydrologic model, the Soil and Water Assessment Tool (SWAT), was applied for hydrologic cycle assessment in the Cong Watershed, Vietnam. After model-output and accuracy-based sensitivity analyses were conducted without and with observation data, respectively, model parameters were calibrated and validated using meteorological and runoff data observed between 1961 and 1975. Model performance in annual, monthly, and daily flow simulations was assessed using the Nash-Sutcliffe Efficiency coefficient and the coefficient of determination (R²). As a result, the accuracy-based sensitivity analysis better identified the key parameters, as compared to the model output-based sensitivity analysis, resulting in better flow simulations in the Cong River. The calibrated model performed well in hydrologic cycle simulations in the Cong Watershed, suggesting the applicability of the SWAT model.

Estimation of water quality dynamics under long-term anoxic state in organically polluted reservoir by field observations and improved ecosystem model

In closed water bodies with significant organic pollution, anoxification due to thermal stratification leads to the elution of nitrogen and phosphorus from the bottom sediment and the generation of sulfide, resulting in further degradation of the water environment. This study focuses on the water quality dynamics in an organically polluted reservoir exhibiting long-term anoxification using two approaches: (1) field observations of seasonal changes in vertical profiles of dissolved oxygen, nitrogen, phosphorus, and sulfide and (2) construction of a water quality prediction model based on an ecosystem model incorporated with anaerobic biochemical processes. Iron and sulfate reduction occurred simultaneously because nitrate–nitrogen was reduced by denitrifying bacteria after the anoxification, and iron reduction became the main factor of the increase in ammonium–nitrogen and phosphate–phosphorus. The redox state of the bottom sediment surface, when anoxification began to occur, greatly affected the water quality dynamics caused by gradual reductive reactions under anaerobic conditions. Furthermore, the calculation accuracy of ammonium–nitrogen, phosphate–phosphorus, and sulfide was highly improved by modifying the conventional model based on the field observations. The characteristics of water quality under anaerobic conditions were sufficiently reflected in the upgraded ecosystem model. The proposed water quality prediction model could be used to quantitatively estimate the water environment dynamics in organically polluted water bodies. The model could be developed further in the future to solve the problems caused by long-term anoxification.

Assessment of Depth Measurement Using an Acoustic Doppler Current Profiler and a CTD Profiler in a Small River in Japan

Recent advances in acoustic Doppler current profiler ADCP allow for measuring spatially continuous and high resolution hydraulic data such as water depth and flow velocity that are the basis for ecohydraulic analysis and modelling. In this paper, an ADCP and a conductivity-temperature-depth CTD profiler were used for depth measurement in Zuibaji river in Fukuoka, Japan, and accuracies of these measurements were assessed and compared based on tape-measured water depth with a GPS coordinate. As a result, water depth measured with a CTD profiler and an ADCP showed a good agreement with observation data, which supports the applicability of the ADCP and CTD for bathymetric survey. A major source of error seems to be a positioning error for obtaining GPS coordinates on site. Further study considering measurement errors, systematic errors and positioning errors, may be needed for a deeper understanding of error characteristics, leading to a better application of these innovative technologies for ecohydraulic surveys.