Average Salinity Of The Ocean

Measuring Salinity Without Wasting Your Day

The number you will find in most textbooks is thirty-five parts per thousand. That is a rounded global mean, useful for rough calculations but not for anything that requires precision. When I was running conductivity-temperature-depth surveys on a small research vessel in the North Atlantic, we were trying to map a freshwater plume near a river outflow. The CTD sensor was reading fine, but the conductivity cell had a minor biofouling issue. Our practical salinity readings drifted by about two percent over six hours. That sounds small until you are trying to track a specific water mass boundary and every hundred meters looks like random noise. What actually happened was the biofilm changed the effective surface area of the conductivity electrodes. I cleaned the cell with a soft brush and distilled water rinse between casts, which dropped the drift back down to under one percent. We logged the pre- and post-cast calibration values and applied a linear correction during post-processing. It saved us from having to discard three days of good casts because the data looked suspicious.

What the Average Salinity Of The Ocean Actually Means

Practical salinity is dimensionless. It is calculated from the ratio of the electrical conductivity of a seawater sample to the conductivity of a standard potassium chloride solution, adjusted for temperature and pressure. The old units of grams per kilogram or parts per thousand are still widely used in older literature and by people who do not work with oceanography regularly. One practical salinity unit is roughly equivalent to one gram of dissolved salt per kilogram of seawater, but they are not exactly the same thing. The distinction matters if you are doing mass-balance calculations or mixing studies. The global mean sits somewhere between thirty-four and thirty-five PSU across the open ocean. Regional variation is the real story. The Mediterranean Sea runs around thirty-eight PSU because evaporation outpaces precipitation and inflow from the Atlantic. The Baltic Sea can drop below seven PSU in the deep basins. The Red Sea hits nearly forty PSU in the summer. These numbers are not theoretical. They are measured repeatedly with calibrated instruments and show up in every Atlantic or Pacific transect you will look at. Salinity changes with depth in predictable patterns in most places. Surface water is fresher where there is rain or ice melt. It gets saltier where evaporation dominates. Below the mixed layer, you usually hit a halocline where salinity changes rapidly over a few hundred meters. Below that, in the deep ocean, salinity is relatively uniform. The North Atlantic Deep Water forming near Greenland sits around thirty-five PSU. Antarctic Bottom Water is slightly fresher, closer to thirty-four point six PSU. These differences are small but significant for density calculations and circulation models.

How to Work With Salinity Data Without Making Stupid Mistakes

First, always check whether your source uses practical salinity, salinity on the PSS-78 scale, or the newer EOS-2008 standard. Most older papers and datasets use PSS-78 practical salinity. The difference is usually tiny, on the order of a few hundredths of a PSU, but it adds up when you are combining datasets from different eras or different research groups. I once spent two hours debugging a density profile that looked wrong before I realized the salinity values had been converted using the wrong formula. The dataset itself was fine. If you are measuring salinity yourself in the field, don't trust the onboard sensor without verifying it against a lab reference. The in situ instruments drift. Temperature compensation helps but does not eliminate it. A simple way to check is to take a water sample, bring it back to the ship's lab, and measure it with a calibrated laboratory refractometer or a titration. The two methods should agree within about zero point zero two PSU for modern equipment. If they don't, something is wrong with the sensor, the calibration, or the sample handling. The refractometer method is fast enough that I usually run it every three to four hours during a deployment. Another thing people overlook is the effect of suspended sediment on conductivity measurements. In turbid coastal waters, particulate matter can interfere with the conductivity cell. I worked on a project in the Gulf of Mexico where the Mississippi River plume was dumping huge amounts of sediment into the estuary. The CTD salinity readings were consistently two to three percent lower than the lab measurements. The sediment increased the bulk conductivity of the water independently of the dissolved ions. We solved it by filtering a subset of samples through a one micron filter before lab analysis and applying a correction factor derived from the comparison. The correction varied with location and season, so we recalculated it every time we returned to the area.

For most practical purposes, if you need a single representative number for the open ocean, thirty-five PSU is fine. If you need accuracy for a specific location or application, you need to measure it or pull it from a dataset that matches your conditions. World Ocean Atlas data is freely available from NOAA and provides monthly, seasonal, and annual climatological fields at various resolutions. For a quick look at a specific coordinate, you can download the nearest grid point and get salinity values with vertical resolution down to fifty meter intervals in the upper ocean. It won't replace actual measurements but it is better than guessing. There is also the question of how salinity interacts with temperature in density calculations. Seawater density depends on both, and the relationship is nonlinear. A small change in salinity at high pressure and low temperature has a different effect on density than the same change at the surface. If you are doing any calculation involving water mass identification or mixing ratios, you need to use a proper equation of state, not a simple linear approximation. The TEOS-10 standard handles this correctly and is what most modern oceanographers use. Older programs still relying on the UNESCO equation of state will give slightly different results, usually in the third or fourth decimal place of potential density. That may or may not matter depending on what you are doing.

When Salinity Data Is Completely Useless

I will be straightforward about this because it is a common source of wasted effort. UnCalibrated or poorly documented salinity data can be worse than no data at all. I have seen datasets where the salinity values were recorded without temperature information, or where the temperature was logged in Celsius and the salinity sensor was reporting raw conductivity without any conversion. Processing these retroactively is possible in some cases but requires assumptions that introduce uncertainty. If the original metadata is missing, you are often stuck. Another scenario where salinity data falls apart is in polar regions during winter. Sea ice formation rejects salt, which increases the salinity of the underlying water. But the process is complicated by brine entrapment, frost flower formation, and the fact that the water below the ice is often in a state of convective overturn. Simple point measurements at a single depth during a brief ship visit can miss the entire vertical structure. What you measure at the surface might be fresh from recent snowmelt while ten meters down the water is significantly saltier. Without a full vertical profile, the single number you report could be misleading. And finally, satellite-derived salinity has improved a lot over the last decade, but it is still a surface measurement with limited depth penetration. The SMOS and SMAP missions give you estimates of sea surface salinity, usually within about one PSU of in situ values in the open ocean. That is useful for large-scale circulation studies. It is not useful if you need to know what is happening below the first few meters, especially in regions with strong stratification or fresh surface layers. Don't try to use satellite salinity as a substitute for profiling data unless you are only interested in broad spatial patterns.

If you want the current best estimate for global average surface salinity, the open ocean mean is approximately thirty-four point six PSU according to recent Argo float compilations. The number shifts slightly depending on which years and which processing version you use. The important thing is that it is an average, and averages hide the variation that actually matters for most applications.