Supraj Gunda ’27, Computer Science

Supraj Gunda ’27

Major

Computer Science

Project Title

A Comparative Analysis of Oceanographic Variables as Predictors of Coral Bleaching Using Machine Learning

Minor(s): Quantitative and Computational Biology, Statistics and Machine Learning

The contemporary understanding of coral bleaching is that corals expel their symbiotic algae mainly due to rising ocean temperatures, which eventually leads to their death. However, studies like those produced by the Deutsch research group propose that corals may be facing an oxygen deficit (hypoxia) that is just as concerning as rising temperature. My project aims to elucidate the puzzling mechanisms behind coral bleaching by determining which oceanographic variables, such as chlorophyll-a, sea surface temperature and salinity levels, most accurately predict coral bleaching severity. My first objective was to synthesize and process oceanographic data from satellite datasets based on timestamps and severity indices from insitu coral bleaching observations. In the second part of my project, I passed combinations of the data to various machine learning models, including Random Forest Classifiers, RNNs and Transformers. On top of diving deep into the major threat of coral bleaching, this experience has taught me to work with big data that spans oceans and decades. Most of all, this project has bolstered my continued interest in the interdisciplinary applications of computer science, by showing how my local Python script can impact the understanding of coral reefs the world over.

* This internship is connected to the HMEI Biodiversity Grand Challenges project, “Assessing the Costs of Biodiversity Loss in Hypoxic Coral Reef Ecosystems.”