Article
Environmental Sciences
Qiqiao Huang, Sheng Chen, Jinkai Tan
Summary: This study proposes a deep learning-based convolutional neural network called TSRC for precipitation nowcasting in China. TSRC compensates for current local cues with previous local cues during convolution processes to retain more contextual information and reduce uncertain features. Experimental results show that TSRC outperforms OF and UNet models in terms of forecasting performance, with higher probability of detection, lower false alarm rate, smaller mean absolute error, and higher structural similarity index, especially at longer lead times. Furthermore, TSRC exhibits better performance in forecasting high-intensity radar echoes, indicating its potential for improving precipitation intensity forecasting. Future research will focus on the combination of multi-source data and further improvements in model architecture.
Article
Meteorology & Atmospheric Sciences
Xian Xiao, Juanzhen Sun, Xiushu Qie, Zhuming Ying, Lei Ji, Mingxuan Chen, Lina Zhang
Summary: The study proposes and implements a proof-of-concept method for assimilating total lightning observations in the 4DVAR framework. By assimilating both radar and lightning data simultaneously, improved dynamical consistency, enhanced updraft and latent heat, and improved moisture distributions are achieved. The combined data assimilation scheme proves to be robust to variations in vertical velocity profiles, radii of horizontal interpolation, binning time intervals, and relationships used to estimate maximum vertical velocity from lightning flash rates.
MONTHLY WEATHER REVIEW
(2021)
Article
Economics
Saulius Jokubaitis, Dmitrij Celov, Remigijus Leipus
Summary: This study examines the use of sparse methods to forecast the expenditure components of the US and EU GDP in the short-run before official data release. The results show that sparse methods can outperform benchmark methods and improve forecast accuracy.
INTERNATIONAL JOURNAL OF FORECASTING
(2021)
Article
Multidisciplinary Sciences
Mariel Abigail Cruz-Najera, Mayra Guadalupe Trevino-Berrones, Mirna Patricia Ponce-Flores, Jesus David Teran-Villanueva, Jose Antonio Castan-Rocha, Salvador Ibarra-Martinez, Alejandro Santiago, Julio Laria-Menchaca
Summary: This paper addresses the problem of forecasting real-life crime and compares four simple and four machine-learning-based ensemble forecasting methods. It also proposes five forecasting techniques to handle the seasonal component of the time series. The results show that the simple moving average with seasonal removal techniques perform the best for these series.
Article
Engineering, Civil
Rajesh Maddu, Indranil Pradhan, Ebrahim Ahmadisharaf, Shailesh Kumar Singh, Rehana Shaik
Summary: This study explores the relevance of large-scale climate phenomenon indices in improving short-term reservoir inflow prediction. A framework combining machine learning algorithms and climate variables is developed, and an ensemble model is created using a weighted voting method. The model consistently outperforms standalone algorithms in predicting high and low flows in two different reservoirs.
JOURNAL OF HYDROLOGY
(2022)
Article
Thermodynamics
Fei Wang, Shuang Tong, Yiqian Sun, Yongsheng Xie, Zhao Zhen, Guoqing Li, Chunmei Cao, Neven Duic, Dagui Liu
Summary: This paper proposes an ultra-short-term wind speed hybrid prediction method based on wind process pattern forecasting. By dividing the wind process into different patterns and selecting the corresponding prediction model based on the pattern, the proposed method can reliably forecast future wind speeds.
Article
Environmental Sciences
Deepak Gupta, Narayanan Natarajan, Mohanadhas Berlin
Summary: Wind energy is a potential renewable energy source globally. Accurate prediction of wind speed is crucial for estimating wind power accurately. Hybrid machine learning models were used in this study for short-term wind speed prediction, with LDMR model outperforming others in prediction accuracy and ELM model being computationally faster.
ENVIRONMENTAL SCIENCE AND POLLUTION RESEARCH
(2022)
Article
Materials Science, Multidisciplinary
Masataka Mizuno, Kazuki Sugita, Hideki Araki
Summary: Using first principles-based Monte Carlo simulations, we investigated the short-range order (SRO) in CrMnFeCoNi high-entropy alloy. The results showed a significant change in the distribution of nearest neighbor (NN) pairs, leading to the formation of L1(2)-type ordering. This chemical SRO is caused by the instability of parallel-spin NN pairs around Cr atoms. The formation of this SRO is predicted to increase the elastic modulus of the material.
RESULTS IN PHYSICS
(2022)
Article
Energy & Fuels
Kamini Shahare, Arghya Mitra, Dipanshu Naware, Ritesh Keshri, H. M. Suryawanshi
Summary: This paper explores different stochastic and deterministic approaches for short-term load forecasting by collecting two years' worth of historical load data. Traditional methods and machine learning methods are compared, with the CNN-LSTM hybrid model being selected as the best method with a correlation coefficient of 95.05%.
Article
Geosciences, Multidisciplinary
Aurelie Bouchard, Magalie Buguet, Adrien Chan-Hon-Tong, Jean Dezert, Philippe Lalande
Summary: This study focuses on estimating the risk of lightning strikes caused by thunderstorms over the sea within a short-term forecast. Three methods are developed and compared, based on thresholds and weighting functions, a neural network approach, and belief functions. Each method is applied to a dataset and evaluated using a ground truth dataset based on lightning stroke locations. The choice of method depends on the balance between false alarms, missed detections, and runtimes. The first method has a low missed detection rate but a high false alarm rate, while the other two methods have lower false alarm rates but higher missed detection rates. The third method is the fastest among the three.
Article
Environmental Sciences
Jinhui He, Hao Yang, Shijie Zhou, Jing Chen, Min Chen
Summary: In this study, a dual-attention mechanism multi-channel convolutional LSTM (DACLSTM) model was proposed for wind speed prediction. The model utilized European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5) near-ground element-grid data and selected elements highly correlated with wind speed to form multiple channels. Experimental results demonstrated that the DACLSTM model outperformed the traditional ConvLSTM model and fully connected network long short-term memory (FC_LSTM) in predicting wind speed with a lead time of six hours.
Article
Geosciences, Multidisciplinary
Maryse Charpentier-Noyer, Daniela Peredo, Axelle Fleury, Hugo Marchal, Francois Bouttier, Eric Gaume, Pierre Nicolle, Olivier Payrastre, Maria-Helena Ramos
Summary: This paper presents a methodological framework for the event-based evaluation of short-range hydrometeorological ensemble forecasts in flash-flood events. The framework focuses on anticipating and accurately localizing discharge threshold exceedances. The proposed approach includes evaluating rainfall forecasts, analyzing the flood rising limb at ungauged sub-catchments, and evaluating forecast hydrographs at selected gauged sub-catchments. The framework is tested for a flash flood event in France and evaluated three ensemble rainfall nowcasting research products.
NATURAL HAZARDS AND EARTH SYSTEM SCIENCES
(2023)
Article
Meteorology & Atmospheric Sciences
J. K. Fletcher, C. A. Diop, E. Adefisan, M. A. Ahiataku, S. O. Ansah, C. E. Birch, H. L. Burns, S. J. Clarke, J. Gacheru, T. D. James, C. K. Ngetich Tuikong, D. Koros, V. S. Indasi, B. L. Lamptey, K. A. Lawal, D. J. Parker, A. J. Roberts, T. H. M. Stein, E. Visman, J. Warner, B. J. Woodhams, L. H. Youds, V. O. Ajayi, E. N. Bosire, C. Cafaro, C. A. T. Camara, B. Chanzu, C. Dione, W. Gitau, D. Groves, J. Groves, P. G. Hill, I. Ishiyaku, C. M. Klein, J. H. Marsham, B. K. Mutai, P. N. Ndiaye, M. Osei, T. I. Popoola, J. Talib, C. M. Taylor, D. Walker
Summary: Testbeds have played a crucial role in advancing weather forecasting worldwide. The African Science for Weather Information and Forecasting Techniques (SWIFT) program recently conducted the first high-impact weather testbed in tropical Africa, involving researchers and forecasters from multiple African countries, the United Kingdom, and international organizations. The testbed focused on trialing new forecasting and nowcasting products, engaging users and researchers, and generating feedback for future research and development. The outcomes of the testbed, including improved forecasts and recommended operating procedures, have strengthened partnerships and garnered support from funding agencies and organizational directors.
BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY
(2023)
Article
Automation & Control Systems
Md Alamgir Hossain, Evan Gray, Junwei Lu, Md Rabiul Islam, Md Shafiul Alam, Ripon Chakrabortty, Hemanshu Roy Pota
Summary: This article proposes a novel framework, CEMOLS, to improve the prediction accuracy of very short-term wind power generation. The framework combines CEEMDAN, MBO, and LSTM models and demonstrates an improvement in forecasting accuracy compared to the benchmark model.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Energy & Fuels
Shreya Sajid, Surender Reddy Salkuti, C. Praneetha, K. Nisha
Summary: This paper focuses on short-term wind speed forecasting using time series methods. Various time series forecasting techniques are applied and compared using performance metrics. A novel LSTM-ARIMA model is proposed, which achieves the highest prediction accuracy and the least error metrics at all time scales.
ENERGY SOURCES PART A-RECOVERY UTILIZATION AND ENVIRONMENTAL EFFECTS
(2022)