This is a python port of the mixed layer depth algorithm outlined in Holte and Talley 2009.
To install from the pypi repository:
pip install holteandtalley
If you would like to install it from source in order to make changes:
#run this in the cloned repository
pip install -e .
this will allow you to import holteandtalley from anywhere on your machine, but any changes you make to the source will be immediately reflected.
from holteandtalley import HolteAndTalley
h = HolteAndTalley(pressures,temperaturess,salinities,densities)
##The temperature algorithms mixed layer depth
h.tempMLD
##The salinity algorithms mixed layer depth
h.salinityMLD
##The density algorithms mixed layer depth
h.densityMLD
It is also possible to use only the temperature algorithm
from holteandtalley import HolteAndTalley
h = HolteAndTalley(pressures,temperatures)
##The temperature algorithms mixed layer depth
h.tempMLD
While I tried to make this as close to the Matlab code provided in the supplementary materials of the paper I was unable to make it exactly the same. Below is a table which shows the max,mean, and standard deviation of difference between the output of the matlab library and this one after being run on the 138 profiles provided in the supplementary materials of the original paper.
| Algorithm | Max Difference | Mean Difference | Standard Deviation Difference |
|---|---|---|---|
| Temperature | 0.5 dbar | 0.06 dbar | 0.13 dbar |
| Density | 12 dbar | 0.275 dbar | 1.447 dbar |
| Salinity | 8.0 dbar | 0.12 dbar | 0.8 dbar |
NOTE: The version of the matlab library I used I slightly modified to correct for some quirks of the library which I think are bugs.
The original paper does not include flowcharts for the density and salinity algorithms so I have included those below transcribed from the matlab supplementary materials provided.
The MLTFit is defined as the interection of the line of best fit of the thermocline and the mixed layer. The best fit line of the thermocline is the line tangent to the profile at the point where d(T)/d(P) is the greatest. To determine the best fit line of the mixed layer, the algorithm calculates the best fit over progressively more points in the profile, calculates the error over the points included in the fit and then finds the deepest fit line which satisfies an error tolerance ET < 10-10.
The deepest point in the profile which is less than 0.2C compared to the SST
T(iMLTFIT)-T(iMLTFIT+2)
25 dbar
The minimum of the gradient maximum and the temperature maximum if that is less than 100 dbar. Otherwise it is set to 0
Deepest point such that the gradient is less than 0.005 C/dbar or max if that criteria is not met.
