Estimating sea temperature trends using a linear Gaussian state-space model in Jogashima, Kanagawa, Japan

Abstract

The quantitative evaluation of the trend of seawater temperature is of significant importance, and the development of a statistical model that accurately reflects real data and can be readily estimated represents a key objective. In this study, we used a linear Gaussian state-space model to estimate the level components representing the trend, taking into account the seasonality, the influence of the Kuroshio, and changes in the trend, for seawater temperatures observed over a 25-year period in Jogashima, Kanagawa Prefecture. The results showed that the seawater temperature exhibited an increase when the distance from Nozimazaki to the Kuroshio axis is close, and during the fall and winter months when the Kuroshio is A type. The incorporation of Kuroshio data into the model resulted in enhanced precision in seawater temperature predictions. The trend of seawater temperature exhibited a downward trajectory until 2007, after which it exhibited an upward trajectory. When the trend was estimated using a simple linear regression model, the model residuals exhibited significant autocorrelation and did not satisfy the underlying assumptions of the linear regression model. It would be prudent to exercise caution when applying simple linear regression models and interpreting the results of seawater temperature trend analysis.