Abstract Deciphering freshwater processes in the Bay of Bengal (BoB) is paramount for understanding ocean‐atmospheric coupled processes in the Northern Indian Ocean. Characterizing thermohaline structures during freshening is essential for quantifying mixing processes and reducing salinity biases in ocean models. Despite the increase in observations, traditional property‐based approaches in isolation are often inadequate to characterize freshwater‐driven thermohaline structures and their variability. To address this, we employed a machine learning approach, the Gaussian Mixture Model (GMM), to classify fresher profiles (Sea Surface Salinity ≤32 psu) and their associated temperature structures using observations in the BoB. Our analysis identified five distinct Hyposalinity classes, each with unique thermohaline characteristics. Monsoon‐driven freshwater input modulates the seasonal thermohaline evolution of the BoB, resulting in three distinct thermohaline zones. During June‐August, freshening initiates in the northern and northeastern BoB (Zones A & B), producing strong salinity stratification. In the northern region, freshwater input and cyclonic circulation maintain a shallow mixed layer. The interplay of Ekman transport, and planetary waves deepens the isothermal layer in the northeastern region. Near surface freshwater from these zones are advected into the central BoB (Zone C) during winter, where surface cooling promotes temperature inversions. From February‐April, prevailing anticyclonic circulation in the BoB enhances vertical mixing, leading to a transition in the waters toward higher salinity. By May, the freshwater initiated during the previous June is thoroughly mixed and completes the BoB freshwater cycle.
Sasidharan et al. (Wed,) studied this question.