←Back to World Data Ocean
•2 min read•from Frontiers in Marine Science | New and Recent Articles

Short-term and long-term prediction of South China Sea SST based on multiple meteorological factors and machine learning

Our take

This study presents a machine learning-driven multivariate framework for predicting sea surface temperature (SST) in the South China Sea, integrating key meteorological factors to enhance forecasting accuracy. Unlike traditional univariate models, this approach utilizes multiple variables, including wind components, dew point temperature, and cloud cover, to capture complex, nonlinear relationships. Evaluating models such as Random Forest, XGBoost, and LightGBM, the research demonstrates that the Random Forest model achieves the highest accuracy for both short-term and long-term forecasts, emphasizing the importance of a multivariate input design.
Short-term and long-term prediction of South China Sea SST based on multiple meteorological factors and machine learning

The recent study on the short-term and long-term prediction of sea surface temperature (SST) in the South China Sea highlights a critical advancement in our understanding of climate dynamics. By utilizing a machine learning-driven multivariate framework, the research integrates essential meteorological variables to unveil complex, nonlinear relationships that traditional univariate models often overlook. This approach signifies a paradigm shift in SST forecasting, which is crucial given that sea surface temperatures directly influence global climate patterns and ecological stability. Such innovations align with the broader need for strategic investment in the ocean economy, as discussed in articles like World Economic Forum: Here's why we need Strategic investment in the Ocean economy. and Beneath the waves, the ocean holds a hidden record of our planet’s changing climate..

The study's incorporation of seven meteorological and hydrological factors—including wind components, dew point temperature, and cloud cover—reveals the intricate interplay between atmospheric conditions and SST. Notably, the research indicates that total cloud cover plays a more significant role in SST predictions than sea surface salinity, shedding light on what factors should be prioritized in future climate models. Furthermore, the Random Forest model's superior accuracy for both short-term and long-term forecasts emphasizes the potential of machine learning technologies in enhancing our predictive capabilities. As climate change accelerates, the ability to forecast SST with improved accuracy could provide invaluable insights for policymakers and environmentalists alike.

This advancement matters not just for academic circles but for a broad audience concerned about climate change's impacts on marine ecosystems. The South China Sea, a region of immense biodiversity and economic significance, is particularly vulnerable to shifts in SST. As highlighted in the article Islands of biodiversity created by remote Arctic kelp forests of the central Kitikmeot Sea, the health of ocean ecosystems is intricately tied to temperature variations. Thus, the findings from this study could inform conservation strategies and sustainable practices in marine resource management, emphasizing the need for action grounded in scientific understanding.

Looking ahead, the integration of machine learning in climate science raises important questions about the future of SST predictions and their implications for global climate resilience. As we continue to refine our models and methodologies, we must consider how these advancements can be leveraged to foster international collaboration in ocean stewardship. The ability to predict SST with a lead time of 20 months opens avenues for proactive rather than reactive strategies in addressing climate change impacts. As researchers and practitioners build upon these findings, the challenge will be to ensure that the knowledge gained translates into actionable policies that prioritize the health of our oceans and the communities that depend on them.

In conclusion, as we advance our scientific understanding and predictive capabilities, the question remains: How can we best translate these insights into effective global action that addresses the urgent challenges facing our oceans? The ocean's future—and our own—may depend on the answers we find in the data.

Sea surface temperature (SST) is a vital component of the climate system, and its spatiotemporal variations significantly influence global climate and ecological equilibrium. Unlike most existing univariate SST prediction models that neglect atmospheric forcing information, this study proposes a machine learning-driven multivariate framework integrating key meteorological variables to learn data-driven, nonlinear relationships for improved SST prediction. Based on the ERA5 reanalysis data, three machine learning algorithms, namely, Random Forest (RF), XGBoost and LightGBM, are used to construct short-term and long-term SST forecast models for the South China Sea. The input feature variables include seven meteorological and hydrological variables such as SST, 10m u-component of wind (U10), 10m v-component of wind (V10), 2m dewpoint temperature (d2m), 2m temperature (t2m), mean sea level pressure (SLP), and total cloud cover (TCC). Correlation analysis revealed that these meteorological factors are significantly correlated with SST, with the strongest correlations observed for 2-meter dew point temperature and 2-meter air temperature. Model performance is assessed using metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R²). The results indicate that the RF model exhibits the highest accuracy for both short-term and long-term forecasting models. Furthermore, this study explores high-resolution SST forecasting models for the South China Sea, revealing that total cloud cover (TCC) contributes more to SST predictions than sea surface salinity (SSS), and the model performs well across most areas of the South China Sea (excluding coastal regions), achieving forecasts with a lead time of at least 20 months. The long lead-time prediction ability derived from a multivariate input design further highlights the advantages of the proposed method over traditional single-factor models. These findings demonstrate the feasibility of machine learning algorithms for SST prediction, providing an efficient approach to understanding future SST changes and their potential impacts, while emphasizing the necessity of integrating multiple meteorological factors to enhance forecasting accuracy.

Read on the original site

Open the publisher's page for the full experience

View original article →

Tagged with

#climate change impact