Modelling the seasonal cycle of dissolved oxygen in the upper ocean at ocean weather station P

Abstract

Three main processes regulate the variations of dissolved O 2 concentrations in the surface waters: gas exchange at the air-sea interface, vertical mixing and biological activity of marine organisms. A one-dimensional integral mixed layer model ( Gaspar, 1988) is used to study the temporal evolution of monthly averaged dissolved O 2 content of surface waters at Ocean Weather Station P, and to assess the relative importance of the various contributing mechanisms during 1969–1972. Production and consumption due to biological activity are taken into account as an input function of the model. A large part of the seasonal signal of dissolved O 2 in surface waters can be reproduced by the physical model without biological activity. However, kinetics of gas exchange, biological production and entrainment of sub-mixed layer water all contribute by the same order of magnitude to supersaturation during warming periods and undersaturation during cooling periods. Various shapes (over depth and time) of production-consumption function have been tested for the year 1970. Most of the evolution of monthly average dissolved O 2 in the surface waters can be obtained (1) with a total annual production rate of the order of 5 mol O 2 m −2 y −1, (2) with a constant production throughout the year and in the 0–50 m layer, and (3) with logarithmic decrease in consumption between 50 and 300 m. The relative influence of various parameters on the three processes supplying O 2 to the surface waters is investigated. Total annual production P seems to be the most influential. Vertical mixing and depth of photic zone, z 0, affect the gas exchange flux during the cooling season. Episodic events, like storms, modify the supersaturation of the mixed layer O 2 content by up to 4 mmol m −3, but gas exchange later draws back this content towards a smooth evolution curve. Finally, the sensitivity of the net annual gas exchange to various parameters is too large for the model to provide a reliable value.