Analysis of Global Research Trends in Artificial Intelligence for Meteorology Using LDA Topic Modeling (Focus on Papers from 2018 to 2024)

LDA 토픽 모델링을 활용한 글로벌 기상 분야의 인공지능 연구동향 분석 (2018~2024년 논문 주제 중심)

This study aims to analyze research trends related to Artificial Intelligence (AI) in the global meteorological field from 2018 to 2024 and identify the major research topics and keywords. By utilizing the Web of Science database, a total of 5,846 papers related to AI in meteorology were identified and analyzed. The study employed latent Dirichlet allocation (LDA) topic modeling to extract the main research topics. The optimization of topic modeling parameters was performed by adjusting document-topic density (alpha) and word-topic density (beta) distributions, which control the concentration of topics in documents and words in topics, respectively. Through comprehensive parameter optimization, the model achieved the coherence score of 0.639 with alpha value of 0.08, beta value of 0.01, and 6 topics, indicating clear and well-separated research themes in the field. These optimal parameter values were used for the topic modeling analysis. The analysis revealed that (1) research on 'AI-based prediction of hydrological variables' encompasses studies applying AI techniques to predict hydrological variables such as rainfall and evaporation, aiming for more precise meteorological forecasting. (2) Studies on 'AI-based analysis of the impacts of climate change' utilize AI models to analyze the effects of climate change on various regions and ecosystems, assessing potential impacts under different climate change scenarios and predicting future environmental changes. (3) Research on 'AI-based prediction of oceanic and surface temperatures' focuses on improving the accuracy of meteorological and environmental observations by predicting ocean and land surface temperatures using satellite data. (4) Studies on 'Machine learning-based risk assessment and prediction of natural disasters' evaluate and predict the likelihood of natural disasters such as floods and landslides, providing crucial information for disaster management and prevention. (5) Research on 'AI and meteorological data utilization for real-time rainfall prediction' aims to enhance the accuracy of real-time rainfall forecasting by combining meteorological radar data with AI techniques, playing a critical role in rapidly changing weather conditions. (6) Studies on 'AI utilization in wind power forecasting and meteorological condition analysis' aim to optimize wind energy production by predicting wind speed and weather conditions, contributing to efficient energy management. This study systematically analyzes research trends related to the application of AI in meteorology, contributing to the academic development of the field and suggesting future research directions. Specifically, by identifying research trends through topic modeling, this study provides a structured understanding of the convergence of meteorology and AI, offering valuable foundational data to researchers in the field.