Global Ocean Salinity Measurements Have Some Serious Issues After 2015

Abstract

Ocean salinity is essential for understanding changes in the climate system. Various gridded ocean salinity products, which are largely based on Argo measurements since the 2000s, have been created and used for climate-related studies. However, after 2015 a significant and unrealistic global salinification trend appears in most of the widely used global salinity products and disagreements between those products increase, both of which should be a concern for the climate community.

Global Mean Sea Level Rise Inferred From Ocean Salinity and Temperature Changes

“…These reconstructions include the Chinese Institute of Atmospheric Physics (IAP) product (Wang et al, 2017), the UK Met Office EN4 product (Good et al, 2013), and the Scripps Institution of Oceanography (SIO) product (Roemmich & Gilson, 2009). The IAP and EN4 salinity reconstructions do not exhibit a decreasing trend in globally averaged salinity over the 2001-2019 period, due mainly to an increase in global salinity after 2015 when a known bias (Liu et al, 2022;Ponte et al, 2021) in Argo salinity measurements may affect their results. The SIO reconstruction (Roemmich & Gilson, 2009), which is available from 2004 onwards and relies solely on Argo float data, does not show a substantial increase in salinity after 2015.…”

“…Our salinity mapping approach utilizes quality controlled historical observations from 2001 to 2019 consisting of individual salinity casts from the World Ocean Database 2018 (Boyer et al, 2018), including ocean station data (OSD), conductivity-temperature-depth (CTD) profiles, autonomous profiling floats (PFL, primarily Argo floats), and autonomous pinniped bathythermographs (APB) (Text S1 in Supporting Information S1). Because a number of Argo floats suffer from a high salinity bias starting in 2015 (Liu et al, 2022;Ponte et al, 2021;Wang et al, 2017;Wong et al, 2023), we produced a global bias correction for Argo float data post-2015 akin to what has previously been applied to biased temperature probes (Levitus et al, 2009; Text S2 in Supporting Meltwater from grounded ice (a) such as glaciers and continental ice sheets directly contributes to GMSLR by adding mass to the ocean, known as barystatic sea level rise (SLR), as freshwater flows from the land to the sea. This meltwater also dilutes the ocean, causing salinity to drop.…”

“…However, previous attempts to quantify barystatic SLR using ocean freshening trends have been hampered by the small signal of ocean freshening relative to the large natural variability of regional ocean salinity, insufficient sampling coverage, and instrumental biases (Liu et al., 2022; Llovel et al., 2019; Ponte et al., 2021; Wang et al., 2017). Because of these issues, global ocean mass budgets constructed from ocean salinity measurements show unrealistically large interannual to decadal variability and uncertain long‐term trends (Liu et al., 2022; Llovel et al., 2019; Munk, 2003; Ponte et al., 2021; Wang et al., 2017). Therefore, barystatic SLR has typically been inferred from the difference between the observed SLR and the thermosteric SLR (Llovel et al., 2019), by remote sensing of changes in ocean mass by satellite gravity measurements (Watkins et al., 2015), or by remote sensing of mass losses from glaciers and ice sheets (Bamber et al., 2018).…”

“…The barystatic component can in principle be derived from measurements of seawater salinity, since barystatic SLR is driven by the input of freshwater into the ocean which reduces ocean salinity (Munk, 2003) (Figure 1). However, previous attempts to quantify barystatic SLR using ocean freshening trends have been hampered by the small signal of ocean freshening relative to the large natural variability of regional ocean salinity, insufficient sampling coverage, and instrumental biases (Liu et al., 2022; Llovel et al., 2019; Ponte et al., 2021; Wang et al., 2017). Because of these issues, global ocean mass budgets constructed from ocean salinity measurements show unrealistically large interannual to decadal variability and uncertain long‐term trends (Liu et al., 2022; Llovel et al., 2019; Munk, 2003; Ponte et al., 2021; Wang et al., 2017).…”

“…The autoregressive approach places realistic constraints on the temporal evolution of global ocean salinity, reducing uncertainties in ocean freshening trends by a factor of four compared to previous studies (Liu et al., 2022; Ponte et al., 2021). This method is validated against output from several Earth system models, and is also demonstrated to be robust to known instrumental biases that affect salinity measurements of some Argo autonomous profiling floats since 2015 (Liu et al., 2022; Ponte et al., 2021) (Section 2 and Supporting Information ). In Section 3, the ARANN global salinity reconstructions are compared with previous salinity reconstructions, and are also evaluated against existing estimates of the ice budget, ocean mass change, and GMSLR from satellite altimeters.…”