Dependency Investigation of Sea Surface Temperature on Sea Bottom Temperature and Sea Surface Salinity

Sea Surface Temperature is the most popular parameter under analysis to sense the variations in Sea. Better sensors clubbed with still better recording instruments, over the period of time, have made analysis easier for the researchers to investigate the dependencies amongst various parameters. The proposed work is an attempt in the same direction. Analysis of data collected over almost a century, dating from 1916 to 2014 at the site La Jolla, West California, USA by staff from Scripps Institution of Oceanography reveals significant relationship between Sea Surface Temperature, Sea Surface Salinity and Sea Bottom Temperature. A feed forward neural network using Back Propagation Algorithm is used to predict the future values by optimizing the count of delay and hidden neurons. Using these optimum values, additional exogenous input is added to the network and the error is computed. It is observed that the performance of the system with exogenous inputs gets better and the error reduces. Multiple step prediction for the same data is also performed and the errors are tabulated in terms of MSE, RMSE and NRMSE.