SULI 2021: Sea Surface Temperature Forecasting

For the summer of 2020, I participated in the Science Undergraduate Laboratory Internship (SULI) at Pacific Northwest National Lab (PNNL). My project was to explore deep-learning techniques for long-term sea-surface temperature (SST) forecasting.

In the resulting manuscript, coauthored with my internship mentors, we explore various autoencoder architectures inspired by Koopman Operator Theory. We find that for this particular problem, the architecture that made the least assumptions about the internal state of the autoencoder’s latent space provided the best results.

All opinions expressed are my own and do not reflect those of PNNL or other collaborators.