Recent regime of persistent hypoxia in the Baltic Sea

Abstract

Oxygen deficiency, in the form of hypoxia and anoxia, is a direct consequence of the eutrophication of the Baltic Sea. There is ongoing debate concerning the increasing extent of hypoxia. We analyse the integral metrics of hypoxia and anoxia: (1) temporal evolution of the hypoxic and anoxic area and volume, (2) the spatial distribution of the probabilities of hypoxia and anoxia occurrence in the Baltic Sea. The results are based on the state-of-the-art coupled physical and biogeochemical numerical model reanalysis data from Copernicus Marine Environment Monitoring Service for the period of 1993–2017. Statistical analysis showed that the variability of hypoxic and anoxic areas since the year 2000 represents stationary processes around their respective mean levels. From 2000 to 2017, the hypoxic area varies between 50000 and 80000 km2 and the anoxic area varies between 10000 and 50000 km2. Different methods and data sources indicate that the uncertainties of the estimates account for about 10000 km2. We suggest that the loss of stationarity of the time series of the hypoxic area would be an indication of the regime change of hypoxia development in the Baltic Sea. Probability distribution maps of hypoxia and anoxia provide detailed information about the persistency of hypoxia and anoxia in different parts of the Baltic Sea. The probability of hypoxia exceeds 0.9 in the eastern and western Gotland basins and in the deep area of the Bornholm basin. The Gulf of Finland and the shallower areas that connect different deep basins of the Baltic Sea exhibit seasonal and episodic hypoxia. The 80 m and 120 m isobaths are the approximate bathymetry limits of hypoxia and anoxia occurrence, respectively. Our study supports previous knowledge that hypoxia development is controlled to a large degree by the depth of the permanent halocline.

A Regime Shift Toward a More Anoxic Environment in a Eutrophic Sea in Northern Europe

“…A possible regime shift in the levels of anoxic area extent in the Baltic proper have previously been reported by Hansson et al (2011) and Kõuts et al (2021) to occur in 1999. In the present study, we added observation data from another 9 years to the analysis by Hansson et al (2011) and included data prior to 1993 compared to the model study by Kõuts et al (2021). According to the analysis of our longer time series a regime shift in the anoxic area extent occurred in 2001.…”

“…We therefore estimate the temporal change in extent of anoxic and hypoxic areas in the Baltic proper. Partly, this has previously been studied using re-analyses 3Dmodel data (Kõuts et al, 2021). However, in the present study we divide the Baltic proper into sub-basins and use longer time-series of observational data.…”

Copernicus Marine Service Ocean State Report, Issue 5

“…The area of the Baltic Sea bottom has been the main subject for the investigations of increasing hypoxia and anoxia (Carstensen et al 2014;Reusch et al 2018;Meier et al 2019). The volume prone to low oxygen concentration (≤2 ml/l) envelopes ∼2500 km 3 of water (Kõuts et al 2021). The Gulf of Bothnia, which has well-oxygenated water due to weak vertical stratification, consists of a large water volume of 2865 km 3 below the depth of 80 m. Therefore, taking into consideration the total volume of the Baltic Sea, the decrease of the oxygen pool is seemingly moderate.…”

Investigating Hypoxic and Euxinic Area Changes Based on Various Datasets From the Baltic Sea

“…Uncertainties related to data quality were estimated by comparing the hypoxic and euxinic areas defined by the IOW data with those described in a mapping product of the SMHI (Hansson and Andersson, 2015;Hansson and Viktorsson, 2020), in the observational dataset ALM by Almroth-Rosell et al (2021) and in two model-based three-dimensional ocean reanalyses, i.e., the Rossby Centre Ocean model (RCO) reanalysis by Liu et al (2017) and the Copernicus reanalysis (Reana) for the Baltic Sea by Axell et al (2019) and Liu et al ( 2019, see also Kõuts et al, 2021).…”

“…Why the two datasets, ALM and Reana, significantly differ in their long-term trends since 1998 is unknown. For further details about the trends in Reana the reader is referred to Kõuts et al (2021).…”