sea surface temperature prediction using artificial neural networks

In this study we create the following models: ARIMA, MLP, LSTM, CNN, LSTM-CNN. They are trained /tested by ANN in order to predict the future value of SST for the period of observation (2003 to 2020) over the regions with 10 locations. The models are validated by RMSE. Then the SST data were analyzed for seasonal and annual variations.

Anna University | April 2021

Undergraduate Final Year Thesis Project

Team: A. Kamali, M. Sabeena, Pooja Gopi (Team Lead)

Guide: Dr. S. Jayalakshmi

Sea Surface Temperature timelapse of the raw data downloaded from the MODIS Sensor from the years 2003 to 2020 over each month

Poster of the Project

Sea Surface Temperature pattern over the years from 2003 to 2020 for each point

Analysis: Created a code in Python using Tensorflow and Keras libraries for prediction, Performed data cleaning for NetCDF gridded files using NCO packages