The study presents the SG-MCMC method for analyzing large spatial datasets, focusing on ocean temperature data from the Argo Program. By deploying the SGRLD method, the authors effectively model mean and covariance functions, tracking temperature variations at various depths. The model was compared with the NNGP method using prediction metrics such as MSE and squared correlation. The results indicate that SGRLD efficiently handles the dataset, providing valuable insights into ocean temperature variations over time, while maintaining computational feasibility.
The proposed SG-MCMC method is applied to analyze ocean temperature data, demonstrating its efficacy against competing methods like NNGP, especially in large spatial datasets.
Using the Argo Program's ocean temperature data, we utilize the SGRLD method to efficiently analyze mean and covariance functions, achieving favorable prediction metrics.
Collection
[
|
...
]