Search

Abstract In river research, forecasting flow velocity accurately in vegetated channels is a significant challenge. The forecasting performance of various independent and hybrid machine learning (ML) models are thus quantified for the first time in this work. Utilizing flow velocity measurements in both natural and laboratory flume experiments, we assess the efficacy of four distinct standalone machine learning techniques—Kstar, M5P, reduced error pruning tree (REPT) and random forest (RF) models. In addition, we also test for eight types of hybrid ML algorithms trained with an Additive Regression (AR) and Bagging (BA) (AR-Kstar, AR-M5P, AR-REPT, AR-RF, BA-Kstar, BA-M5P, BA-REPT and BA-RF). Findings from a comparison of their predictive capabilities, along with a sensitivity analysis of the influencing factors, indicated: (1) Vegetation height emerged as the most sensitive parameter for determining the flow velocity; (2) all ML models displayed outperforming empirical equations; (3) nearly all ML algorithms worked optimal when the model was built using all of the input parameters. Overall, the findings showed that hybrid ML algorithms outperform regular ML algorithms and empirical equations at forecasting flow velocity. AR-M5P (R2 = 0.954, R = 0.977, NSE = 0.954, MAE = 0.042, MSE = 0.003, and PBias = 1.466) turned out to be the optimal model for forecasting of flow velocity in vegetated-rivers.

Water quality assessment and prediction play crucial roles in ensuring the sustainability and safety of freshwater resources. This study aims to enhance water quality assessment and prediction by integrating advanced machine learning models with XAI techniques. Traditional methods, such as the water quality index, often require extensive data collection and laboratory analysis, making them resource-intensive. The weighted arithmetic water quality index is employed alongside machine learning models, specifically RF, LightGBM, and XGBoost, to predict water quality. The models' performance was evaluated using metrics such as MAE, RMSE, R2, and R. The results demonstrated high predictive accuracy, with XGBoost showing the best performance (R2 = 0.992, R = 0.996, MAE = 0.825, and RMSE = 1.381). Additionally, SHAP were used to interpret the model's predictions, revealing that COD and BOD are the most influential factors in determining water quality, while electrical conductivity, chloride, and nitrate had minimal impact. High dissolved oxygen levels were associated with lower water quality index, indicative of excellent water quality, while pH consistently influenced predictions. The findings suggest that the proposed approach offers a reliable and interpretable method for water quality prediction, which can significantly benefit water specialists and decision-makers.

Abstract Soil temperatures at both surface and various depths are important in changing environments to understand the biological, chemical, and physical properties of soil. This is essential in reaching food sustainability. However, most of the developing regions across the globe face difficulty in establishing solid data measurements and records due to poor instrumentation and many other unavoidable reasons such as natural disasters like droughts, floods, and cyclones. Therefore, an accurate prediction model would fix these difficulties. Uzbekistan is one of the countries that is concerned about climate change due to its arid climate. Therefore, for the first time, this research presents an integrated model to predict soil temperature levels at the surface and 10 cm depth based on climatic factors in Nukus, Uzbekistan. Eight machine learning models were trained in order to understand the best-performing model based on widely used performance indicators. Long Short-Term Memory (LSTM) model performed in accurate predictions of soil temperature levels at 10 cm depth. More importantly, the models developed here can predict temperature levels at 10 cm depth with the measured climatic data and predicted surface soil temperature levels. The model can predict soil temperature at 10 cm depth without any ground soil temperature measurements. The developed model can be effectively used in planning applications in reaching sustainability in food production in arid areas like Nukus, Uzbekistan.

Machine learning techniques have garnered considerable attention in modern technologies due to their promising outcomes across various domains. This paper presents the comprehensive methodology of an optimized and efficient forecasting approach for Particulate Matter 10, specifically tailored to predefined locations. The execution of a comparative analysis involving eight models enables the identification of the most suitable model that aligns with the primary research objective. Notably, the test results underscore the superior performance of an ensemble model, which integrates state-of-the-art methodologies, surpassing the performance of the other seven state-of-the-art models. Adopting a case-specific methodology with machine learning techniques contributes to achieving a notably high regression coefficient (R²≈1) across all models. Furthermore, the study underscores the potential for future endeavors in predicting location-specific environmental factors. • This study focused on forecasting PM10 with machine learning models with the consideration of air quality factors and meteorological factors • Ensemble model was developed for the forecasting purposes with higher performance.

Rice is a crucial crop in Sri Lanka, influencing both its agricultural and economic landscapes. This study delves into the complex interplay between economic indicators and rice production, aiming to uncover correlations and build prediction models using machine learning techniques. The dataset, spanning from 1960 to 2020, includes key economic variables such as GDP, inflation rate, manufacturing output, population, population growth rate, imports, arable land area, military expenditure, and rice production. The study's findings reveal the significant influence of economic factors on rice production in Sri Lanka. Machine learning models, including Linear Regression, Support Vector Machines, Ensemble methods, and Gaussian Process Regression, demonstrate strong predictive accuracy in forecasting rice production based on economic indicators. These results underscore the importance of economic indicators in shaping rice production outcomes and highlight the potential of machine learning in predicting agricultural trends. The study suggests avenues for future research, such as exploring regional variations and refining models based on ongoing data collection.

The Rainfall-Runoff (R-R) relationship is essential to the hydrological cycle. Sophisticated hydrological models can accurately investigate R-R relationships; however, they require many data. Therefore, machine learning and soft computing techniques have taken the attention in the environment of limited hydrological, meteorological, and geological data. The accuracy of such models depends on the various parameters, including the quality of inputs and outputs and the used algorithms. However, identifying a perfect algorithm is still challenging. This study develops a fuzzy logic-based algorithm called Cascaded-ANFIS to accurately predict runoff based on rainfall. The model was compared against three regression algorithms: Long Short-Term Memory, Grated Recurrent Unit, and Recurrent Neural Networks. These algorithms have been selected due to their outstanding performances in similar studies. The models were tested on the Mahaweli River, the longest in Sri Lanka. The results showcase that the Cascaded-ANFIS-based model outperforms the other algorithms. The correlation coefficient of each algorithm’s predictions was 0.9330, 0.9120, 0.9133, 0.8915, 0.6811, 0.6811, and 0.6734 for the Cascaded-ANFIS, LSTM, GRU, RNN, Linear, Ridge, and Lasso regression models respectively. Hence, this study concludes that the proposed algorithm is 21% more accurate than the second-best LSTM algorithm. In addition, Shared Socio-economic Pathways (SSP2-4.5 and SSP5-8.5 scenarios) were used to generate future rainfalls, forecast the near-future and mid-future water levels, and identify potential flood events. The future forecasting results indicate a decrease in flood events and magnitudes in both SSP2-4.5 and SSP5-8.5 scenarios. Furthermore, the SSP5-8.5 scenario shows drought weather from May to August yearly. The results of this study can effectively be used to manage and control water resources and mitigate flood damages.

Hydrologic models to simulate river flows are computationally costly. In addition to the precipitation and other meteorological time series, catchment characteristics, including soil data, land use, land cover, and roughness, are essential in most hydrologic models. The unavailability of these data series challenged the accuracy of simulations. However, recent advances in soft computing techniques offer better approaches and solutions at less computational complexity. These require a minimum amount of data, while they reach higher accuracies depending on the quality of data sets. The Gradient Boosting Algorithms and Adaptive Network-based Fuzzy Inference System (ANFIS) are two such systems that can be used in simulating river flows based on the catchment rainfall. In this paper, the computational capabilities of these two systems were tested in simulated river flows by developing the prediction models for Malwathu Oya in Sri Lanka. The simulated flows were then compared with the ground-measured river flows for accuracy. Correlation of coefficient (R), Per cent-Bias (bias), Nash Sutcliffe Model efficiency (NSE), Mean Absolute Relative Error (MARE), Kling-Gupta Efficiency (KGE), and Root mean square error (RMSE) were used as the comparative indices between Gradient Boosting Algorithms and Adaptive Network-based Fuzzy Inference Systems. Results of the study showcased that both systems can simulate river flows as a function of catchment rainfalls; however, the Cat gradient Boosting algorithm (CatBoost) has a computational edge over the Adaptive Network Based Fuzzy Inference System (ANFIS). The CatBoost algorithm outperformed other algorithms used in this study, with the best correlation score for the testing dataset having 0.9934. The extreme gradient boosting (XGBoost), Light gradient boosting (LightGBM), and Ensemble models scored 0.9283, 0.9253, and 0.9109, respectively. However, more applications should be investigated for sound conclusions.

The rapidly changing climate affects an extensive spectrum of human-centered environments. The food industry is one of the affected industries due to rapid climate change. Rice is a staple food and an important cultural key point for Japanese people. As Japan is a country in which natural disasters continuously occur, using aged seeds for cultivation has become a regular practice. It is a well-known truth that seed quality and age highly impact germination rate and successful cultivation. However, a considerable research gap exists in the identification of seeds according to age. Hence, this study aims to implement a machine-learning model to identify Japanese rice seeds according to their age. Since agewise datasets are unavailable in the literature, this research implements a novel rice seed dataset with six rice varieties and three age variations. The rice seed dataset was created using a combination of RGB images. Image features were extracted using six feature descriptors. The proposed algorithm used in this study is called Cascaded-ANFIS. A novel structure for this algorithm is proposed in this work, combining several gradient-boosting algorithms such as XGBoost, CatBoost, and LightGBM. The classification was conducted in two steps. First, the seed variety was identified. Then, the age was predicted. As a result, seven classification models were implemented. The performance of the proposed algorithm was evaluated against 13 state-of-the-art algorithms. Overall, the proposed algorithm has a higher accuracy, precision, recall, and F1-score than the others. For the classification of variety, the proposed algorithm scored 0.7697, 0.7949, 0.7707, and 0.7862, respectively. The results of this study confirm that the proposed algorithm can be employed in the successful age classification of seeds.

Machine learning has already been proven as a powerful state-of-the-art technique for many non-linear applications, including environmental changes and climate predictions. Wetlands are among some of the most challenging and complex ecosystems for water level predictions. Wetland water level prediction is vital, as wetlands have their own permissible water levels. Exceeding these water levels can cause flooding and other severe environmental damage. On the other hand, the biodiversity of the wetlands is threatened by the sudden fluctuation of water levels. Hence, early prediction of water levels benefits in mitigating most of such environmental damage. However, monitoring and predicting the water levels in wetlands worldwide have been limited owing to various constraints. This study presents the first-ever application of deep machine-learning techniques (deep neural networks) to predict the water level in an urban wetland in Sri Lanka located in its capital. Moreover, for the first time in water level prediction, it investigates two types of relationships: the traditional relationship between water levels and environmental factors, including temperature, humidity, wind speed, and evaporation, and the temporal relationship between daily water levels. Two types of low load artificial neural networks (ANNs) were developed and employed to analyze two relationships which are feed forward neural networks (FFNN) and long short-term memory (LSTM) neural networks, to conduct the comparison on an unbiased common ground. The LSTM has outperformed FFNN and confirmed that the temporal relationship is much more robust in predicting wetland water levels than the traditional relationship. Further, the study identified interesting relationships between prediction accuracy, data volume, ANN type, and degree of information extraction embedded in wetland data. The LSTM neural networks (NN) has achieved substantial performance, including R2 of 0.8786, mean squared error (MSE) of 0.0004, and mean absolute error (MAE) of 0.0155 compared to existing studies.

Planning and decision-making are critical managerial functions involving the brain’s executive functions. However, little is known about the effect of cerebral activity during long-time learning while planning and decision-making. This study investigated the impact of planning and decision-making processes in long-time learning, focusing on a cerebral activity before and after learning. The methodology of this study involves the Tower of Hanoi (ToH) to investigate executive functions related to the learning process. Generally, ToH is used to measure baseline performance, learning rate, offline learning (following overnight retention), and transfer. However, this study performs experiments on long-time learning effects for ToH solving. The participants were involved in learning the task over seven weeks. Learning progress was evaluated based on improvement in performance and correlations with the learning curve. All participants showed a significant improvement in planning and decision-making over seven weeks of time duration. Brain activation results from fMRI showed a statistically significant decrease in the activation degree in the dorsolateral prefrontal cortex, parietal lobe, inferior frontal gyrus, and premotor cortex between before and after learning. Our pilot study showed that updating information and shifting issue rules were found in the frontal lobe. Through monitoring performance, we can describe the effect of long-time learning initiated at the frontal lobe and then convert it to a task execution function by analyzing the frontal lobe maps. This process can be observed by comparing the learning curve and the fMRI maps. It was also clear that the degree of activation tends to decrease with the number of tasks, such as through the mid-phase and the end-phase of training. The elucidation of this structure is closely related to decision-making in human behavior, where brain dynamics differ between “thinking and behavior” during complex thinking in the early stages of training and instantaneous “thinking and behavior” after sufficient training. Since this is related to human learning, elucidating these mechanisms will allow the construction of a brain function map model that can be used universally for all training tasks.

Automated fruit identification is always challenging due to its complex nature. Usually, the fruit types and sub-types are location-dependent; thus, manual fruit categorization is also still a challenging problem. Literature showcases several recent studies incorporating the Convolutional Neural Network-based algorithms (VGG16, Inception V3, MobileNet, and ResNet18) to classify the Fruit-360 dataset. However, none of them are comprehensive and have not been utilized for the total 131 fruit classes. In addition, the computational efficiency was not the best in these models. A novel, robust but comprehensive study is presented here in identifying and predicting the whole Fruit-360 dataset, including 131 fruit classes with 90,483 sample images. An algorithm based on the Cascaded Adaptive Network-based Fuzzy Inference System (Cascaded-ANFIS) was effectively utilized to achieve the research gap. Color Structure, Region Shape, Edge Histogram, Column Layout, Gray-Level Co-Occurrence Matrix, Scale-Invariant Feature Transform, Speeded Up Robust Features, Histogram of Oriented Gradients, and Oriented FAST and rotated BRIEF features are used in this study as the features descriptors in identifying fruit images. The algorithm was validated using two methods: iterations and confusion matrix. The results showcase that the proposed method gives a relative accuracy of 98.36%. The Fruit-360 dataset is unbalanced; therefore, the weighted precision, recall, and FScore were calculated as 0.9843, 0.9841, and 0.9840, respectively. In addition, the developed system was tested and compared against the literature-found state-of-the-art algorithms for the purpose. Comparison studies present the acceptability of the newly developed algorithm handling the whole Fruit-360 dataset and achieving high computational efficiency.

Hydropower stands as a crucial source of power in the current world, and there is a vast range of benefits of forecasting power generation for the future. This paper focuses on the significance of climate change on the future representation of the Samanalawewa Reservoir Hydropower Project using an architecture of the Cascaded ANFIS algorithm. Moreover, we assess the capacity of the novel Cascaded ANFIS algorithm for handling regression problems and compare the results with the state-of-art regression models. The inputs to this system were the rainfall data of selected weather stations inside the catchment. The future rainfalls were generated using Global Climate Models at RCP4.5 and RCP8.5 and corrected for their biases. The Cascaded ANFIS algorithm was selected to handle this regression problem by comparing the best algorithm among the state-of-the-art regression models, such as RNN, LSTM, and GRU. The Cascaded ANFIS could forecast the power generation with a minimum error of 1.01, whereas the second-best algorithm, GRU, scored a 6.5 error rate. The predictions were carried out for the near-future and mid-future and compared against the previous work. The results clearly show the algorithm can predict power generation’s variation with rainfall with a slight error rate. This research can be utilized in numerous areas for hydropower development.

Codebook-based feature encodings are a standard framework for image recognition issues. A codebook is usually constructed by clusterings, such as the k-means and the Gaussian Mixture Model (GMM). A codebook size is an important factor to decide the trade-off between recognition performance and computational complexity and a traditional framework has the disadvantage to image recognition issues when a large codebook; the number of unique clusters becomes smaller than a designated codebook size because some clusters converge to close positions. This paper focusses on the disadvantage from a perspective of the distribution of prior probabilities and presents a clustering framework including two objectives that are alternated to the k-means and the GMM. Our approach is first evaluated with synthetic clustering datasets to analyze a difference to traditional clustering. In the experiment section, although our approach alternated to the k-means generates similar results to the k-means results, our approach is able to finely tune clusters for our objective. Our approach alternated to the GMM significantly improves our objective and constructs intuitively appropriate clusters, especially for huge and complicatedly distributed samples. In the experiment on image recognition issues, two state-of-the-art encodings, the Fisher Vector (FV) using the GMM and the Vector of Locally Aggregated Descriptors (VLAD) using the k-means, are evaluated with two publicly available image datasets, the Birds and the Butterflies. For the results of the VLAD with our approach, the recognition performances tend to be worse compared to the original VLAD results. On the other hand, the FV using our approach is able to improve the performance, especially in a larger codebook size.