Forecasting Approach Research Articles

Single-instant spatial wave height forecast using machine learning: An image-to-image translation approach based on generative adversarial networks

An innovative single-instant approach for spatial significant wave height forecast based on a Generative Adversarial Networks (GANs) model is presented. The model's most distinctive feature is its ability to provide spatial wave height forecast with only a single snapshot of wind field input, eliminating the need for continuous wind field sequences while simultaneously maintaining high accuracy and enhanced efficiency. The proposed model architecture comprises an image-to-image translation generator and a patch-style discriminator, working in tandem to forecast high-quality wave height data. The model is trained using a 20-year ECMWF Reanalysis v5 (ERA5) wind and wave dataset. A vast sea area covering 15°-40°N, 105°-130°E in West Pacific is used for this study. The presented model is carefully tested against a 10-year period historical dataset and the results are discussed. The model exhibits outstanding overall performances with Mean Absolute Error (MAE) of 0.16 m, Root Mean Square Error (RMSE) of 0.24 m, and Mean Absolute Percentage Error (MAPE) of 13.4 %. The model proves particularly effective in forecasting extreme wave heights during tropical cyclones, capturing the structure of typhoon-driven waves and providing wave heights with favorable accuracy. It is also proficient at delivering satisfactory statistical wave height forecasts for long-term periods.

Advancing short-term solar irradiance forecasting accuracy through a hybrid deep learning approach with Bayesian optimization

The optimization of solar energy integration into the power grid relies heavily on accurate forecasting of solar irradiance. In this study, a new approach for short-term solar irradiance forecasting is introduced. This method combines Bayesian Optimized Attention-Dilated Long Short-Term Memory and Savitzky-Golay filtering. The methodology is implemented to analyze data obtained from a solar irradiance probe situated in Douala, Cameroon. Initially, the unprocessed data is augmented by integrating distinctive solar irradiation variables, and the Savitzky-Golay filter with Bayesian Optimization is used to enhance its quality. Subsequently, multiple deep learning models, including Long Short-Term Memory, Bidirectional Long Short-Term Memory, Artificial Neural Networks, Bidirectional Long Short-Term Memory with Additive Attention Mechanism, and Bidirectional Long Short-Term Memory with Additive Attention Mechanism and Dilated Convolutional layers, are trained and evaluated. Out of all the models considered, the proposed approach, which combines the attention mechanism and dilated convolutional layers, demonstrates exceptional performance with the best convergence and accuracy in forecasting. Bayesian Optimization is further utilized to fine-tune the polynomial and window size of the Savitzky-Golay filter and optimize the hyperparameters of the deep learning models. The results show a Symmetric Mean Absolute Percentage Error of 0.6564, a Normalized Root Mean Square Error of 0.2250, and a Root Mean Square Error of 22.9445, surpassing previous studies in the literature. Empirical findings highlight the effectiveness of the proposed methodology in enhancing the accuracy of short-term solar irradiance forecasting. This research contributes to the field by introducing novel data pre-processing techniques, a hybrid deep learning architecture, and the development of a benchmark dataset. These advancements benefit both researchers and solar plant managers, improving solar irradiance forecasting capabilities.

Offshore wind power prediction based on two-stage hybrid modeling**

Accurate prediction of offshore wind power generation is essential for efficient power scheduling and grid integration. This study introduces an innovative hybrid forecasting approach to address variability in power output due to different climatic conditions and challenges in measuring offshore wind speeds. The approach consists of two primary phases. First, a gated recurrent unit neural network with an attentional mechanism and a quantile regression model forecast wind speeds between cut-in and rated wind speeds, capturing trends under distinct climatic conditions. Second, wind speeds at nine quantile points are used as inputs to a relevance vector machine model, optimized via a cuckoo search algorithm, to establish the relationship between wind speed and power output. An empirical evaluation on a European offshore wind power dataset validates the approach's effectiveness across various climatic conditions. The two-stage model demonstrates enhanced adaptability, offering more reliable power predictions than conventional methods. Results indicate this hybrid forecasting method is more accurate than traditional techniques, significantly improving offshore wind power prediction performance.

Multivariate federated tree-based forecasting combining resilience and privacy: Application to distributed energy resources

Using spatiotemporal data can increase forecasting accuracy for distributed energy resources such as wind farms, solar power plants, and electric vehicle charging stations. The underlying forecasting models assume that data are shared by the different data owners, which is not always feasible because of privacy and confidentiality constraints. Existing methods for privacy-preserving data-sharing are based mainly on linear or non-linear univariate models. This is quite limiting concerning the potential of non-linear multivariate forecasting approaches, since these models can be computationally more efficient than univariate approaches and can learn non-linear relationships. This paper introduces a new federated forecasting approach implementing a multivariate non-linear learning scheme based on decision trees while enabling fully decentralized computation. Results show higher predictive performance than existing linear or univariate energy forecasting models while preserving privacy.

Chaos theory meets deep learning: A new approach to time series forecasting

We explore the influence and advantages of integrating chaotic systems with deep learning for time series forecasting in this paper. It proposes a novel deep learning method based on the Chen system, which leverages the randomness, sensitivity, and diversity of chaotic mapping to enhance the performance and efficiency of deep learning models. We introduce a deep learning framework that integrates chaotic systems, providing an innovative and effective approach for time series forecasting. The research utilizes three different types of deep learning models as baselines—Long Short-Term Memory, Neural Basis Expansion Analysis, and Transformer—and compares them with their chaotic counterparts to demonstrate the impact of chaotic systems on various deep learning architectures. Experimental validation is conducted on thirteen available time series datasets, assessing the models’ forecasting accuracy, runtime, and resource occupancy. The effectiveness and superiority of the chaotic deep learning method are verified across diverse datasets, including stock markets, electricity, and air quality, showcasing significant improvements over traditional model initialization methods.

Species interactions and eco-evolutionary dynamics of dispersal: the diversity dependence of dispersal.

Dispersal plays a pivotal role in the eco-evolutionary dynamics of spatially structured populations, communities and ecosystems. As an individual-based trait, dispersal is subject to both plasticity and evolution. Its dependence on conditions and context is well understood within single-species metapopulations. However, species do not exist in isolation; they interact locally through various horizontal and vertical interactions. While the significance of species interactions is recognized for species coexistence and food web functioning, our understanding of their influence on regional dynamics, such as their impact on spatial dynamics in metacommunities and meta-food webs, remains limited. Building upon insights from behavioural and community ecology, we aim to elucidate biodiversity as both a driver and an outcome of connectivity. By synthesizing conceptual, theoretical and empirical contributions from global experts in the field, we seek to explore how a more mechanistic understanding of diversity-dispersal relationships influences the distribution of species in spatially and temporally changing environments. Our findings highlight the importance of explicitly considering interspecific interactions as drivers of dispersal, thus reshaping our understanding of fundamental dynamics including species coexistence and the emergent dynamics of metacommunities and meta-ecosystems. We envision that this initiative will pave the way for advanced forecasting approaches to understanding biodiversity dynamics under the pressures of global change. This article is part of the theme issue 'Diversity-dependence of dispersal: interspecific interactions determine spatial dynamics'.

Prospect and sustainability prediction of China's new energy vehicles sales considering temporal and spatial dimensions

The development prospect and sustainability of new energy vehicles (NEVs) are facing numerous challenges under the coupling influence of various factors, which has become a major strategic issue in the automotive industry research within China. Establishing a prediction method to forecast the development trends of NEVs under the complex conditions is of great significance. This paper comprehensively addresses the relevant factors influencing the sales of NEVs over temporal depth and spatial breadth, including random white noise and lag effect resulting from lag operators of influencing factors. A multivariate forecasting approach utilizing the ARIMA model and its optimized model, alongside the GBDT model were adopted to forecast NEV sales in China. The findings demonstrate that: 1) NEVs have exhibited a trend of rapid growth, but are still susceptible to random interference and lag effects caused by influencing factors. Key technology breakthroughs, accessibility of charging infrastructure, and national policy support remain as primary drivers of NEVs development; 2) The ARIM–IO–LSTM model achieved high prediction accuracy with a goodness-of-fit R2 exceeding 90%, outperforming the ARIMA and ARIMA-IO models by 23.586% and 20.790%, respectively, while the ARIMA-IO model improved accuracy by 5.126% compared to ARIMA. The method of model combination and outlier incorporation have proved to be practical solutions for strengthening model accuracy. 3) In the next 3–5 years, the NEV industry will enter a stage characterized by steady progress with an upward trend. Adapting the intensity, scope, and direction of policy support, coupled with achieving technological breakthroughs and enhancing infrastructure construction, are essential for sustaining industrial development and attaining desired objectives.

Assessment of CMIP6 GCMs for selecting a suitable climate model for precipitation projections in Southern Thailand

The selection of General Circulation Models (GCMs) is critical due to computational limitations and underlying uncertainties. This study provides a comprehensive assessment of three bias correction (BC) methods, namely the delta change method (DT), quantile mapping (QM), and empirical quantile mapping (EQM). Utilizing precipitation data from 30 observation stations across Southern Thailand, the evaluation encompasses five CMIP6 GCM models (CAMS-CSM1-0, CanESM5, CNRM-CM6-1, CNRM-ESM2-1, IPSL-CM6A-LR). Evaluation metrics, root mean square error (RMSE), mean absolute error (MAE), Pearson's correlation (r), index of agreement (d), and mean bias error (MBE) are employed to assess BC methods. Evaluation measures suggest that the DT method outperforms EQM and QM, with higher accuracy and lower errors (DT: RMSE = 3.61, MAE = 2.32; EQM: RMSE = 3.82, MAE = 3.70). Taylor diagrams show that CNRM-ESM2-1 has the highest correlation across sites (r = 0.36), albeit with a wider dispersion, while CanESM5 has a more balanced performance. Significant annual precipitation increases are projected for different spans 2021-30, 2061-70, and 2091–2100, particularly under SSP585, which will influence flood risk, water management, and climate adaptation. Future projections with SSP585 continuously projecting a larger probability of increased precipitation. The DT method is recommended for the downscaling of CMIP6 GCMs for precipitation projections in Southern Thailand, recognizing its superior performance. The study's findings provide a foundation for informed decision-making and adaptation planning in southern Thailand, urging policymakers to prioritize climate resilience and adaptation strategies while using a multi-model ensemble approach for robust climate forecasts.

Development of an integrated energy management system for off-grid solar applications with advanced solar forecasting, time-of-use tariffs, and direct load control

Effectively managing and maximizing the integration of renewable energy sources is essential for a sustainable power grid due to the stochastic and intermittent nature of renewable energy generation. This study develops a comprehensive Integrated Energy Management System incorporating supply-demand side management in the form of time-of-use credit, direct load control, and generator control to enhance photovoltaic utilization in off-grid applications. A novel three-step solar energy forecasting approach is proposed in this paper, utilizing low-level data fusion and regression models to predict next-day photovoltaic generation with improved accuracy, and a rule-based decision algorithm is developed to correct forecast errors and manage loads dynamically. A techno-economic analysis covering a 20-year duration is carried out for scenarios with and without the integrated energy management system; three configurations are investigated for supplying an off-grid residential home, including diesel generator, diesel generator/photovoltaic system, and diesel generator/photovoltaic system/integrated energy management system. Results reveal that the hybrid configuration with integrated energy management system achieved 44 % and 46 % reductions in costs and carbon dioxide emissions compared to the diesel generator alone, and 8 % and 9 % compared to the diesel generator/photovoltaic setup respectively. The Integrated Energy Management System further enhanced photovoltaic utilisation rate by over 113 % when compared to the diesel generator/photovoltaic system. Further evaluations include customer behaviour impacts, demonstrating that a fully automated system with 100 % compliance significantly outperforms systems with manual customer control, highlighting the detrimental effect of overrides on the efficiency of direct load control. The flexibility of the Integrated Energy Management System framework allows potential adaptation for on-grid applications, showcasing its utility in diverse operational contexts.

Examining 22 Years of Ambient Seismic Wavefield at Mount St. Helens

Abstract An increase in seismic activity precedes most volcanic eruptions. Whereas event-based forecasting approaches have been successful, some eruptions remain unanticipated, resulting in casualties and damage. Our study leverages the recent advancements in ambient field seismology. We explore features extracted from continuous ambient fields using traditional methods, for example, peak ground velocity, peak ground acceleration, root mean square, root median square, real-time seismic amplitude measurement, and novel methods (displacement seismic amplitude ratio and spectral width). In addition, we explore unsupervised learning of higher order wavelet features using scattering networks. We find that combining all the methods was necessary to disentangle the effects of seismic sources from structural changes at Mount St. Helens. Although the ambient wavefield-based approach does not yield additional or more significant precursory signals than event-based methods at Mount St. Helens, our study demonstrates that the ambient wavefield provides supplementary information, mainly about structural changes and complements traditional methods. The ambient seismic wavefield offers additional insights into long-lasting processes. We find enhanced wave attenuation correlating with geochemical measurements. We interpret this as ongoing structural changes, such as dome growth or the evolution of the volcanic conduit system. On annual and decadal timescales, we interpret seasonal seismic attenuation in the shallow subsurface as groundwater fluctuations, corroborated by observations at the nearby Spirit Lake level. This multimethod approach at Mount St. Helens sheds light on a volcanic system’s underlying dynamics and structure.

Hybrid Approach for Advanced Monitoring and Forecasting of Fouling with Application to an Ethylene Oxide Plant

Hybrid Approach for Advanced Monitoring and Forecasting of Fouling with Application to an Ethylene Oxide Plant

Integrating Technical Indicators and Ensemble Learning for Predicting the Opening Stock Price

Accurately predicting stock prices poses a significant challenge due to the dynamic and complex nature of financial markets. This paper introduces a novel method that combines technical indicators with ensemble learning techniques to effectively forecast opening stock prices. Technical indicators offer valuable insights into market trends and patterns, while ensemble learning methods merge multiple models to enhance predictive precision. The study utilizes various technical indicators such as moving averages, Relative Strength Index (RSI), and Bollinger Bands to capture diverse aspects of market behaviour. Ensemble learning techniques like Random Forest, Gradient Boosting, Support Vector Regressor, and ARIMA model are then employed to consolidate the forecasts from these indicators. The proposed framework is assessed using historical stock market data, and extensive experiments showcase its superior performance compared to individual indicators and traditional forecasting approaches. The findings reveal that integrating technical indicators with ensemble learning leads to a significant improvement in accuracy, with a success rate of 91.45% in predicting opening stock prices, thus providing valuable insights for investors and financial analysts.

Forecasting China's agricultural carbon emissions: A comparative study based on deep learning models

Given the critical urgency to combat the escalating climate crisis and the continuous rise in agricultural carbon emissions (ACE) in China, accurately forecasting their future trends is crucial. This research employs the emission factor method to assess ACE throughout mainland China from 1993 to 2021. To refine our forecasting approach, both statistical and neural network methodologies were utilized to pinpoint key factors influencing ACE. We crafted forecasting models incorporating both deep learning techniques and traditional methods. Notably, the Tree-structured Parzen Estimator Bayesian Optimization (TPEBO) algorithm was applied to optimize Long Short-Term Memory (LSTM) neural networks, culminating in the creation of a superior integrated TPEBO-LSTM model that demonstrated strong performance across various datasets. The forecasting outcomes suggest that ACE in 24 provinces are expected to reach their zenith before 2030, primarily driven by farm operations, as well as livestock and poultry manure management. The result provides a significant forecasting tool for assessing agricultural carbon emissions in different regions, offering insights crucial for targeted mitigation strategies.

Structured multifractal scaling of the principal cryptocurrencies: Examination using a self‐explainable machine learning

AbstractThis paper introduces a novel statistical testing technique known as segmented detrended multifractal fluctuation analysis (SMF‐DFA) to analyze the structured scaling properties of financial returns and predict the long‐term memory of financial markets. The proposed methodology is applied to assess the efficiency of major cryptocurrencies, expanding upon conventional approaches by incorporating different fluctuation regimes identified through a change‐point detection test. A single‐factor model is employed to characterize the endogenous factors influencing scaling behavior, leading to the development of a self‐explanatory machine learning approach for price forecasting. The proposed method is evaluated using daily data from three major cryptocurrencies spanning from April 2017 to December 2022. The analysis aims to determine whether the digital market has experienced significant changes in recent years and assess whether this has resulted in structured multifractal behavior. The study identifies common periods of local scaling among the three prices, with a noticeable decrease in multifractality observed after 2018. Furthermore, complementary tests on shuffled and surrogate data are conducted to explore the distribution, linear correlation, and nonlinear structure, shedding light on the explanation of structured multifractality to some extent. Additionally, prediction experiments based on neural networks fed with multi‐fractionally differentiated data demonstrate the utility of this new self‐explanatory algorithm for decision‐makers and investors seeking more accurate and interpretable forecasts.

Hybridizing traditional and next-generation reservoir computing to accurately and efficiently forecast dynamical systems.

Reservoir computers (RCs) are powerful machine learning architectures for time series prediction. Recently, next generation reservoir computers (NGRCs) have been introduced, offering distinct advantages over RCs, such as reduced computational expense and lower training data requirements. However, NGRCs have their own practical difficulties, including sensitivity to sampling time and type of nonlinearities in the data. Here, we introduce a hybrid RC-NGRC approach for time series forecasting of dynamical systems. We show that our hybrid approach can produce accurate short-term predictions and capture the long-term statistics of chaotic dynamical systems in situations where the RC and NGRC components alone are insufficient, e.g., due to constraints from limited computational resources, sub-optimal hyperparameters, sparsely sampled training data, etc. Under these conditions, we show for multiple model chaotic systems that the hybrid RC-NGRC method with a small reservoir can achieve prediction performance approaching that of a traditional RC with a much larger reservoir, illustrating that the hybrid approach can offer significant gains in computational efficiency over traditional RCs while simultaneously addressing some of the limitations of NGRCs. Our results suggest that the hybrid RC-NGRC approach may be particularly beneficial in cases when computational efficiency is a high priority and an NGRC alone is not adequate.

Deep Learning Approaches for Water Stress Forecasting in Arboriculture Using Time Series of Remote Sensing Images: Comparative Study between ConvLSTM and CNN-LSTM Models

Irrigation is crucial for crop cultivation and productivity. However, traditional methods often waste water and energy due to neglecting soil and crop variations, leading to inefficient water distribution and potential crop water stress. The crop water stress index (CWSI) has become a widely accepted index for assessing plant water status. However, it is necessary to forecast the plant water stress to estimate the quantity of water to irrigate. Deep learning (DL) models for water stress forecasting have gained prominence in irrigation management to address these needs. In this paper, we present a comparative study between two deep learning models, ConvLSTM and CNN-LSTM, for water stress forecasting using remote sensing data. While these DL architectures have been previously proposed and studied in various applications, our novelty lies in studying their effectiveness in the field of water stress forecasting using time series of remote sensing images. The proposed methodology involves meticulous preparation of time series data, where we calculate the crop water stress index (CWSI) using Landsat 8 satellite imagery through Google Earth Engine. Subsequently, we implemented and fine-tuned the hyperparameters of the ConvLSTM and CNN-LSTM models. The same processes of model compilation, optimization of hyperparameters, and model training were applied for the two architectures. A citrus farm in Morocco was chosen as a case study. The analysis of the results reveals that the CNN-LSTM model excels over the ConvLSTM model for long sequences (nine images) with an RMSE of 0.119 and 0.123, respectively, while ConvLSTM provides better results for short sequences (three images) than CNN-LSTM with an RMSE of 0.153 and 0.187, respectively.

Long-Term Regional Influenza-Like-Illness Forecasting Using Exogenous Data.

Disease forecasting is a longstanding problem for the research community, which aims at informing and improving decisions with the best available evidence. Specifically, the interest in respiratory disease forecasting has dramatically increased since the beginning of the coronavirus pandemic, rendering the accurate prediction of influenza-like-illness (ILI) a critical task. Although methods for short-term ILI forecasting and nowcasting have achieved good accuracy, their performance worsens at long-term ILI forecasts. Machine learning models have outperformed conventional forecasting approaches enabling to utilize diverse exogenous data sources, such as social media, internet users' search query logs, and climate data. However, the most recent deep learning ILI forecasting models use only historical occurrence data achieving state-of-the-art results. Inspired by recent deep neural network architectures in time series forecasting, this work proposes the Regional Influenza-Like-Illness Forecasting (ReILIF) method for regional long-term ILI prediction. The proposed architecture takes advantage of diverse exogenous data, that are, meteorological and population data, introducing an efficient intermediate fusion mechanism to combine the different types of information with the aim to capture the variations of ILI from various views. The efficacy of the proposed approach compared to state-of-the-art ILI forecasting methods is confirmed by an extensive experimental study following standard evaluation measures.

Multiagency Ensemble Forecast of Wildfire Air Quality in the United States: Toward Community Consensus of Early Warning

Abstract Wildfires pose increasing risks to human health and properties in North America. Due to large uncertainties in fire emission, transport, and chemical transformation, it remains challenging to accurately predict air quality during wildfire events, hindering our collective capability to issue effective early warnings to protect public health and welfare. Here, we present a new real-time Hazardous Air Quality Ensemble System (HAQES) by leveraging various wildfire smoke forecasts from three U.S. federal agencies (NOAA, NASA, and Navy). Compared to individual models, the HAQES ensemble forecast significantly enhances forecast accuracy. To further enhance forecasting performance, a weighted ensemble forecast approach was introduced and tested. Compared to the unweighted ensemble mean, the multilinear regression weighted ensemble reduced fractional bias by 34% in the major fire regions, false alarm rate by 72%, and increased hit rate by 17%. Finally, we improved the weighted ensemble using quantile regression and weighted regression methods to enhance the forecast of extreme air quality events. The advanced weighted ensemble increased the PM2.5 exceedance hit rate by 55% compared to the ensemble mean. Our findings provide insights into the development of advanced ensemble forecast methods for wildfire air quality, offering a practical way to enhance decision-making support to protect public health.

Evaluation of Severity Index of Road Casualties on Mangawan to Chakghat Section of NH-30 Using Machine Learning Approach

Abstract: Each year, over 1.35 million people die in road accidents, and many more are seriously injured, according to the WHO. Predicting the severity of these accidents is vital for improving prevention and response plans. This paper proposes a new method to predict the severity of road accidents using various factors. This research focuses on road accidents at Sohagi Ghat on NH-30. The study looks at accident types and locations along a 3.67 Kilometer section of steep, challenging terrain. A safety audit was conducted on black spots and it revealed that 324 accidents occurred over five years, making this section particularly dangerous. Special safety measures are needed to reduce the risk in this area. The paper aims at forecasting of upcoming highway accidents through moving average approach, exponential smoothing approach and linear forecasting approach using the previous year’s data sets. With the help of Kaggle Data Directory, precise accident data for visualization is formed and then visualization of road accidents is performed on Python programming language-based Anaconda Navigator’s Jupyter Notebook. While accident severity index of Sohagi Ghat NH-30 is also carried out and some results of machine learning based visualization and some forecasting-based results are driven out and compared with each other. The study calculated accident severity and used Logistic Regression to train a model with 0.85 accuracy value. The precision, recall, and F-1 score of the model were perfect. Injury severity percentages were 10.8% fatal, 48.2% serious, and 41% minor. The Accident Severity Index for the NH-30 Sohagi Ghat location was found to be 0.211.

SolarFlux Predictor: A Novel Deep Learning Approach for Photovoltaic Power Forecasting in South Korea

We present SolarFlux Predictor, a novel deep-learning model designed to revolutionize photovoltaic (PV) power forecasting in South Korea. This model uses a self-attention-based temporal convolutional network (TCN) to process and predict PV outputs with high precision. We perform meticulous data preprocessing to ensure accurate data normalization and outlier rectification, which are vital for reliable PV power data analysis. The TCN layers are crucial for capturing temporal patterns in PV energy data; we complement them with the teacher forcing technique during the training phase to significantly enhance the sequence prediction accuracy. By optimizing hyperparameters with Optuna, we further improve the model’s performance. Our model incorporates multi-head self-attention mechanisms to focus on the most impactful temporal features, thereby improving forecasting accuracy. In validations against datasets from nine regions in South Korea, SolarFlux outperformed conventional methods. The results indicate that SolarFlux is a robust tool for optimizing PV systems’ management and operational efficiency and can contribute to South Korea’s pursuit of sustainable energy solutions.