Study on Correlative Factors of Seawater Temperature Based on Regression Analysis

The seawater temperature is important that affects marine ecosystems and marine life. This paper aimed to analyze the main factors that affect seawater temperature using Bottle Database collected at California Cooperative Oceanic Fisheries Investigations stations. The dataset contained seawater samples collected using bottles. Linear regression was performed using temperature as the dependent variable and depth, salinity, oxygen, and potential density as independent variables. Stepwise methods were used to select variables for regression models. To determine the most influential factor affecting temperature, the random forest was used to rank four independent variables. The plot generated by this analysis showed that potential density strongly influenced seawater temperature, followed by salinity, oxygen, and depth. Then four models were built, including Ordinary Multilinear regression, Linear regression considering the interaction, Linear regression with variable transformation, and Linear regression combining interactions and transformation. The 4th Model, had the highest adjusted R-squared and the lowest prediction sum of squares, making it the most useful model. This model indicated that 4 predictor factors all have significant effects on temperature, and their interactions also play a crucial role. Finally, model 4 can be used to predict water temperature based on depth, salinity, oxygen, and potential density, showing that seawater temperature could be predicted efficiently using these relevant variables. This study provides valuable insights into the main factors affecting seawater temperature and demonstrates the usefulness of linear regression and random forest approaches.