The researchers gauged the amount of carbon dioxide dissolved in the Bay of Bengal using a machine learning model. Credit: S. Priyadarshini
Oceanographers have created detailed machine learning-generated maps to understand how much carbon dioxide the Bay of Bengal absorbs11.
Oceans maintain the balance of atmospheric CO2 levels, by absorbing excess gas produced by human activities. However, this has led to an over-saturation, transforming some parts of the ocean into CO2 sources rather than CO2 sinks.
Using machine learning, researchers led by Apurva Joshi at the Indian National Centre for Ocean Information Services in Hyderabad, wanted to address the lack of knowledge on the role of the Bay of Bengal in sinking carbon. They measured sea-surface partial pressure of CO2 (pCO2), to gauge the amount of CO2 dissolved in the ocean.
Their dataset included sea-surface temperature and salinity, and pCO2 observations, along with other factors like chlorophyll concentration, and mixed layer depth, the depth of the ocean where sunlight can penetrate.
The researchers used the Xtreme Gradient Boosting model since it provided a more accurate depiction of the Bay of Bengal's complex carbonate dynamics, affected by the monsoon system, river inputs, and coastal currents.
They show a detailed variation in the sea-surface pCO2 in the Bay of Bengal over space and time, emphasizing the impact of the east India coastal current and freshwater inputs. Their high-resolution INCOIS-ReML model can provide accurate bio-physical models of the ocean carbon cycle, they say.
The model can inform regional and global carbon cycles and future climate models, the researchers say.
