Machine Learning based vertical diffusivity in EPBL mixing scheme used for ocean surface boundary layer #737
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
updating ML_diffusivity to latest dev/gfdl
Ml diffusivity
replaced the sigma to z coord algorithm used to map neural network output with a simpler and better algorithm that finds the shape function on the hz vertical grid.
some typo corrections
extra unwanted file. deleted.
f_lower is a lower cap on abs_f used inside equation for v_0. A cap is required to avoid singularity. Capping at any value below 1 deg is okay, solution is not sensitive. f_lower was tested for 1 deg and 0.1 deg. SST and MLD did not change.
Merging latest dev/gfdl
updating latest dev/gfdl with ML_diffusivity
deleted unwanted file
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Requesting to add a few subroutines to the EPBL vertical mixing module. These changes enhances the existing vertical diffusivity used in EPBL with machine learning, and makes it constrained on second moment closure. Using symbolic regression and empirical fitting, a shape function ( g(\sigma) ) has been formulated that responds to changes in the surface forcing (Latitude, wind stress, surface buoyancy flux, boundary layer depth). g(\sigma) goes from zero to 1 and its skewness changes as per surface forcing conditions. The velocity scale, v_0, is an approximation that depends on (Coriolis f, ustar, and surface buoyancy flux).
When v_0 is combined with g(\sigma) and multiplied by the energetics based boundary layer depth h, i.e \nu = . g(\sigma) X v_0 X h, we get a diffusivity which is constrained on a second moment closure.
The subroutines are activated by using the flags:
Equation_Discovery_shape = True
Equation_Discovery_velocity = True
Both are required to activate the new diffusivity.
The new subroutines have been tested by running the OM4 configuration (Adcroft et al. 2019) and comparing the simulation against observations (SST from WOA and Mixed Layer Depth from ARGO). The equations are approximating the neural networks. The neural network based diffusivity has been published in Sane et al. 2023 ( https://doi.org/10.1029/2023MS003890 ) and the equations based diffusivity will soon be submitted for a publication.
The changes have also been tested using the scale test.
All the commits can be squashed into one as only the latest is relevant.