Search papers, labs, and topics across Lattice.
This study enhances the Adaptive Next-Generation Reservoir Computing (Adaptive NVAR) framework by integrating Singular Value Decomposition (SVD) to improve high-resolution sea surface temperature (SST) forecasting in the East Sea. By compressing SST fields into a low-dimensional representation, the method effectively captures dominant modes of ocean variability while significantly reducing computational complexity. The results demonstrate that this PCA-enhanced approach consistently outperforms traditional forecasting methods, achieving lower errors across various prediction horizons, making it suitable for real-time applications.
Lower forecasting errors and faster computations make this PCA-enhanced framework a game-changer for real-time sea surface temperature predictions.
Accurate forecasting of sea surface temperature (SST) in regional seas such as the East Sea is crucial for monitoring marine ecosystems, assessing climate risks, managing fisheries, and conducting naval operations. Traditional numerical ocean models provide reliable predictions but are computationally expensive and often unsuitable for real-time forecasting. Many deep learning methods also struggle with high-dimensional spatiotemporal ocean data and experience error accumulation over longer forecasting periods. This study builds on our previously proposed Adaptive Next-Generation Reservoir Computing (Adaptive NVAR) framework, initially introduced and tested on synthetic dynamical systems, and extends it to ocean forecasting. We present a reduced-order forecasting framework that combines Singular Value Decomposition (SVD) with Adaptive NVAR to predict SST dynamics in the East Sea. SST fields are compressed into a low-dimensional representation using SVD, which extracts dominant modes of ocean variability. Adaptive NVAR models the temporal evolution of these latent states, and the predicted states are reconstructed into SST forecasts. We evaluate the framework using regional ocean datasets and compare it with the standard NG-RC/NVAR. Results show that Adaptive NVAR consistently achieves lower forecasting errors across multiple prediction horizons. In addition, SVD reduces computational complexity, resulting in a fast and scalable framework suitable for real-time ocean forecasting.
