Subsurface Hypoxia Observation in the Changjiang Estuary Based on a Wave-Driven Profiler, Satellite Data, and Machine Learning

Abstract

The extent and daily change of coastal hypoxia is difficult to observe and predict, because conventional ship-based observations hardly present spatiotemporal variation of dissolved oxygen (DO) on a day-to-day or shorter timescale. We established a machine learning model between the surface and 15-m depth layer based on high-resolution in situ profile data off the Changjiang estuary (CE). Satellite data were incorporated into this model to obtain spatiotemporal distribution of DO at 15-m depth in the CE at an hourly scale in August 2022. Additionally, we calculated net community production (NCP) within the euphotic zone on a daily scale by using a modified semianalytical model that is suitable for the CE to explore the coupling relationship between 15-m depth DO and NCP. Results show that biological effects dominate DO variation on the nearshore side of the Changjiang diluted water (CDW). Furthermore, we found a time lag of 5 and 4 days for 15-m depth DO behind euphotic zone NCP here, and hypoxia is likely to occur when biological effects reach certain thresholds (NCP > 93.45 mg C m⁻² d⁻¹ in zone A and 216.62 mg C m⁻² d⁻¹ in zone B). Combined effects of CDW and Kuroshio subsurface water (KSSW) not only modulate stratification but also induce upwelling along the steep slope near the offshore zone of CDW, where these complex circulation dynamics create conditions conducive to frequent hypoxia. Our study provides valuable insights into the hypoxia mechanism and development, as well as potential application for operational prediction of hypoxia in coastal waters.