From Wilf Gardner, Texas A & M University, August 8, 1997

Since the New Hampshire meeting [July, 1997], we have compared the MLD
based on the temperature criteria of 0.1 and 0.5 degrees C compared with
the density differences of 0.03 and 0.125 density units.  About 75% of
the time the values are identical.  There are other occasions, however,
where there are significant differences, especially for the 0.1 degrees
C/0.03 comparison. When there is a difference, the temperature
calculations generally give deeper MLDs than density calculations.
Differences most often occur when there is a salinity increase below
the surface.  Temperature criteria are often used for MLDs on moored
data as salinity sensors are sometimes subject to greater drift than
temperature sensors.

None of the above criteria identify the "mixed-layer depth" perfectly in
all cases, but examination of depth plots where there are differences
suggest that the density criteria more reliably identify the depth of most
recent mixing.  For consistency, we recommend that the MLDs based on
density be used when you are trying to calculate mass budgets.  In some
situations, such as sometimes during the TN045 intermonsoon period, there
is no true mixed layer - just a depth at which the MLD criteria are finally
exceeded.  We suggest that you look at an expanded plot of the density
profile if your calculations are critical.