There is no easy way to directly measure the salinity or density of seawater.  However, there are a number of methods to calculate either parameter using accepted equations of state.  Essentially, if any three of the parameters listed below are known, it is possible to mathematically calculate the values for the other three parameters.  The parameters include: 

• temperature,
• pressure,
• conductivity
• salinity,
• sound speed (sound velocity), and
• density.

These equations used to calculate salinity, density and sound velocity - or to reverse calculate conductivity, temperature, or pressure - include:

• Conductivity, temperature and pressure conversion to salinity; (Hill K.D., Dauphinee T.M., Woods D.J.,The Extension of the Practical Salinity Scale 1978 to Low Salinities, IEEE Journal of Oceanic Engineering, Vol. OE-11, No.1, Jan 1986),
• Sound speed, temperature and pressure conversion to salinity; Dakin D.T., Applied Microsystem's Equation for Salinity Calculation from SV,T&P, AML application note, 1999,
• Salinity, temperature and pressure conversion to density; Millero F.J., Poisson A.,International one-atmosphere equation of state of seawater, Deep Sea Research, Vol. 28A, No. 6, pp. 625 to 629, 1981.  Also known as EOS80.
  

As explained above, salinity can be calculated in one of two fashions: 1) using conductivity, temperature and pressure; or 2) using sound velocity, temperature, and pressure.   The following table summarizes salinity accuracies of the two methods, assuming static (non profiling) conditions:

Salinity Specifications by Calculation Approach

	Accuracy (psu)	Precision (psu)	Resolution (psu)	Range (psu)
Salinity from CTD	+/- 0.01 psu	+/- 0.005 psu	+/- 0.001 psu	Up to 40 psu
Salinity from SVTP	+/- 0.035 psu	+/- 0.005 psu	+/- 0.001 psu	Up to 40 psu

Mismatches in sensor response times (primarily between C & T) cause anomalous errors in the salinity data.  These errors are typically referred to as salinity spikes.  Since salinity, temperature and pressure are used to calculate density and sound speed these data spikes also occur in those data sets.  The greater the mismatch in sensor response times between sensors, the larger the salinity spikes will be.  The salinity spike magnitude is also proportional to the thermal and conductivity gradients and the profiling speed.

The three charts on the following page demonstrate salinity spiking. The ‘actual’ and ‘measured’ data were mathematically derived to demonstrate the principle.  The profiles were conducted at a 0.5 m/s descent speed and sampled at 10 Hz.  The solid lines are the ‘actual’ water conditions and the dotted lines are the ‘measured’ and calculated data. 

The two ‘measured’ data sets were derived using typical Applied Microsystems CTD sensor response times.  Our Micro CTD uses a 100ms response temperature sensor and the CTD Plus v2 uses a 350ms response temperature sensor. Both CTD’s use a 25ms response conductivity sensor and a 10ms response pressure sensor. 

The first chart - see below - shows the ‘actual’ and ‘measured’ temperature and conductivity of a ten meter column of water.  Note the depression of the thermocline caused by the slow response time of the temperature sensor.  Also note the smoothing of the small scale structure of the temperature profile.  This is where the majority of the errors originate.

The second chart shows the salinity and density profiles calculated from the ‘measured’ data as well as the ‘actual’ data set.  Note the false indications of less saline water intrusions in the salinity profile.  Also note the depression in depth and stronger picnoclines in the density profile.

In the field the actual data is never present so it is very difficult to assess the magnitude of the errors due to salinity spiking.  In this case, since there is not a large change in salinity the salinity spiking is readily apparent even without looking at the ‘actual’ profile.  However in profiles with strong haloclines the salinity spiking can be difficult to see in a profile, though the errors are still there.

The final chart shows the calculated sound speed profiles.  The salinity profiles are also included for reference.  The depth depression and smoothing effects can be seen in the sound speed profiles.