Skip to content

Minor fixes related to Argo float running#248

Merged
jemrobinson merged 9 commits intomainfrom
247-evaluate-argo-float
Apr 9, 2026
Merged

Minor fixes related to Argo float running#248
jemrobinson merged 9 commits intomainfrom
247-evaluate-argo-float

Conversation

@jemrobinson
Copy link
Copy Markdown
Member

@jemrobinson jemrobinson commented Apr 7, 2026

Question

Should we add Argo float data to the default setup?

Testing method

  • run over sample datasets with or without inclusion of sample Argo data
  • simple model: naive-unet-naive with n_parameters:
    • encoder_sic_icenet: 2
    • encoder_era5: 56
    • encoder_float_argo: 4
    • processor: 11.0M
    • decoder: 32
  • local run on MPS
  • 50 epochs

Results

No Argo (baseline) With Argo Persistence
SIE error 481134 1370787 293761
RMSE 0.07688 0.08967 0.02352
MAE 0.02199 0.01893 0.00320
Train it/s 3.82 3.98 35.35
sie_error rmse mae

No Argo

no-argo

With Argo

with-argo

Persistence

persistence

@jemrobinson jemrobinson requested review from IFenton and louisavz April 7, 2026 10:40
@jemrobinson jemrobinson changed the title Evaluate effect of adding Argo float data Evaluate impact of adding Argo float data Apr 7, 2026
@github-actions
Copy link
Copy Markdown

github-actions Bot commented Apr 7, 2026

Coverage report

Click to see where and how coverage changed

FileStatementsMissingCoverageCoverage
(new stmts)
Lines missing
  icenet_mp/data_processors/filters
  nan_to_num.py 17
  icenet_mp/data_processors/sources
  argo.py 44-45, 188-189, 235-241, 248-253
  ftp.py 66-68
Project Total  

This report was generated by python-coverage-comment-action

@IFenton
Copy link
Copy Markdown
Contributor

IFenton commented Apr 8, 2026

Reviewing this

Copy link
Copy Markdown
Contributor

@IFenton IFenton left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM (though I don't think it closes #247)

@jemrobinson
Copy link
Copy Markdown
Member Author

LGTM (though I don't think it closes #247)

What more would you like to see before we could close #247?

@IFenton
Copy link
Copy Markdown
Contributor

IFenton commented Apr 9, 2026

What more would you like to see before we could close #247?

Sophie and I added some comments to that issue about what it would be good to do to test the impact of adding the Argo float data (e.g. testing assumptions about depth / time window, gridding mechanisms etc.). I don't think they need to be done as part of this PR, but if we're going to close #247 we should capture those comments in a new issue about improving the Argo float data. And I would also update the title of #247 to reflect this (e.g. Evaluate impact of current implementation for adding Argo float data)

Comment thread icenet_mp/config/data/datasets/samp_floatsouth_argo_pt_2017_2019_24h_v1.yaml Outdated
@jemrobinson jemrobinson changed the title Evaluate impact of adding Argo float data Add Argo float data as a default dataset Apr 9, 2026
@jemrobinson jemrobinson changed the title Add Argo float data as a default dataset Minor fixes related to Argo float running Apr 9, 2026
@jemrobinson jemrobinson force-pushed the 247-evaluate-argo-float branch from e938eac to f25a8a2 Compare April 9, 2026 14:02
@jemrobinson
Copy link
Copy Markdown
Member Author

Merging as set of fixes to Argo source but nothing related to changing defaults.

@jemrobinson jemrobinson merged commit 95415f3 into main Apr 9, 2026
6 checks passed
@jemrobinson jemrobinson deleted the 247-evaluate-argo-float branch April 9, 2026 14:28
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants