Prepare addition/replacement of p-model example data by fabern · Pull Request #322 · geco-bern/rsofun

fabern · 2025-11-14T10:34:54Z

This PR

includes data from the rsofun documentation paper as the new example data sets.
- details are summarised in the NEWS.md. briefly,
  - we now use runread_pmodel for both daily and onestep simulation by using a column onestep == TRUE/FALSE in the params_siml of the pmodel_drivers. This replaces the use of run_pmodel_onestep_f_bysite() by the end user.
  - we updated example data sets with additional sites and renamed them to pmodel_drivers and pmodel_validation to reduce potential bugs (also previous data sets were renamed to p_model_oldformat_drivers and p_model_oldformat_validation and will soon be entirely removed from the package)
  - we now support more complex priors as showcased in the documentation paper, see createMixedPrior()
updates the vignettes to a similar calibration as shown in the documentation paper.

Note that, for simplicity, we could have similarly updated the biomee_validation data.frame to a similar format. Note that the drivers biomee_[xxx]_drivers does not need an update (since there is no onestep function for biomee).

For the moment it was decided against updating biomee_validation (in case we want to do later, see code in commit c10f8cc). But such an approach will likely be needed for future multi-target calibrations of BiomeE.

For a summary of changes have a look at NEWS.md

This uses the interface and data format of the target data.frame as used in the rsofun_doc repository. TODO: next steps are - to make this backward compatible with the old input format and arguments, if possible. - - do this by replacing the examples and tests of calib_sofun() with calib_sofun_parallelized() across the package - - and eventually replace the files in data/*.rda - to remove traces of the VJ target - also use get_mod_obs_pmodel_bigD13C_vj_gpp() for predictive plot (time series) in the vignettes.

Since we update the data format, this keeps the "_oldformat" around for development. During development we want: - to replace occurrences of "_oldformat" to the new format - keep around "_oldformat" to ensure backward compatible functions.

…l..' This is to replace the `..._oldformat_...` in upcoming commits.

This is done by removing `onestep` simulations from the new data set and explicitly treating multiple sites when plotting.

* runread_pmodel() had a bug for onestep. Now fixed. * Also includes changes to tests to remove warnings.

This reduces code duplication (parallel and non-parallel versions) and makes pmap more robust by using named arguments instead of relying on positional arguments.

… svg)" This reverts commit 37519c3.

…ample data

…n level of cost functions.

This also updates `cost_likelihood_pmodel()` and`cost_rmse_pmodel()`

…d replace with new cost_likelihood_pmodel1 => cost_likelihood_pmodel_legacy cost_likelihood_pmodel2 => cost_likelihood_pmodel

stineb · 2025-12-16T13:53:31Z

Great work!
Now may be a good time to run styler for tidyverse-styling throughout.
I didn't merge also because some tests are still failing.

fabern · 2025-12-16T13:57:35Z

Yes, the macOS test is currently failing probably due to some network issue installing an unrelated package xfun.
I will run the styler using the non-strict formatting.

fabern · 2025-12-16T14:18:09Z

Okay, I did now: library(styler); styler::style_pkg(style = styler::tidyverse_style, strict=FALSE).
And got the following output:

Styling  34  files:
 R/calib_sofun_legacy.R                           ℹ 
 R/calib_sofun.R                                  ℹ 
 R/cost_likelihood_biomee.R                       ℹ 
 R/cost_likelihood_pmodel_legacy.R                ℹ 
 R/cost_likelihood_pmodel.R                       ℹ 
 R/cost_rmse_biomee.R                             ℹ 
 R/cost_rmse_pmodel.R                             ℹ 
 R/createMixedPrior.R                             ℹ 
 R/data.R                                         ℹ 
 R/luluc.R                                        ℹ 
 R/run_biomee_f_bysite.R                          ℹ 
 R/run_pmodel_f_bysite.R                          ℹ 
 R/run_pmodel_onestep_f_bysite.R                  ℹ 
 R/runread_biomee_f.R                             ⚠ 
 R/runread_pmodel_f.R                             ℹ 
 tests/testthat.R                                 ✔ 
 tests/testthat/helpers.R                         ℹ 
 tests/testthat/test-calibration-biomee.R         ℹ 
 tests/testthat/test-calibration-pmodel.R         ℹ 
 tests/testthat/test-likelihoods.R                ℹ 
 tests/testthat/test-model-runs-snapshot-biomee.R ℹ 
 tests/testthat/test-model-runs.R                 ℹ 
 tests/testthat/test-quantitative-validation.R    ℹ 
 data-raw/generate_biomee_drivers.R               ℹ 
 data-raw/generate_biomee_output.R                ℹ 
 data-raw/generate_pmodel_drivers.R               ℹ 
 data-raw/generate_pmodel_output.R                ℹ 
 vignettes/biomee_luluc.Rmd                       ℹ 
 vignettes/biomee_use.Rmd                         ℹ 
 vignettes/data_format.Rmd                        ℹ 
 vignettes/new_cost_function.Rmd                  ℹ 
 vignettes/param_calib_multitarget.Rmd            ℹ 
 vignettes/pmodel_use.Rmd                         ℹ 
 vignettes/sensitivity_analysis.Rmd               ℹ 
────────────────────────────────────────────────
Status	Count	Legend 
✔ 	1	File unchanged.
ℹ 	32	File changed.
✖ 	1	Styling threw an error.
────────────────────────────────────────────────

I hope this does not bite me for the merge of phydro. (But I think I could separately do the same in the phydro branch to reduce potential issues.)

library(styler); styler::style_pkg(style = styler::tidyverse_style, strict=FALSE)

stineb · 2025-12-16T14:49:53Z

I don't think this will bite you as phydro mainly has new developments at the Fortran level, which is left untouched by styler.

…ings_runMCMC This should help to make code clearer and easier to distinguish between these two settings. And this also add renames `sampler` to `sampler_runMCMC` for consistency. This is fully backwards compatible because of partial name matching for argument names (settings_calibrate) and for list element names (settings_runMCMC, sampler_runMCMC). Nevertheless, for settings_calibrate we still included a warning about deprecated use.

n_chains_independent => nrChains n_parallel_independent => n_parallel_nrChains

fabern

Self-review done. This can be merged!

This reverts commit 448e0e6.

…tes-with-gmd-paper" This reverts commit 3a93439, reversing changes made to b5999ca. This is to ensure installing the package from GitHub and checking the Documentation (geco-bern.github.io/rsofun/) do both refer to v5.1.0 over the holidays.

…gmd-paper" This reverts commit 3a93439, reversing changes made to b5999ca. This is to ensure installing the package from GitHub and checking the Documentation (geco-bern.github.io/rsofun/) do both refer to v5.1.0 over the holidays.

…-gmd-paper" This reverts commit 4849557.

fabern · 2026-01-06T13:58:16Z

This has been reverted in the main branch to avoid overwriting the documentations website.

The main branch has now been changed to distinguish between release and development documentations.
Thus the changes from this reverted PR will be merged again under PR #324.

So this can remain closed.

fabern added 30 commits November 14, 2025 11:33

Prepare addition/replacement of p-model example data

6033eca

Remove analysis subfolder

32ef14a

Add tests for createMixedPrior()

70fdc0b

Update docs and comments

2c7ab6d

Add development notes to NEWS.md

62fc99e

Name new data format for drivers and validation (observations) 'pmode…

f1aa32b

…l..' This is to replace the `..._oldformat_...` in upcoming commits.

Make old pmodel_use.Rmd work with new data set

156fec9

This is done by removing `onestep` simulations from the new data set and explicitly treating multiple sites when plotting.

fix R Cmd check by updating documentation

9f62000

Make runread_pmodel() compatible with updated inputs (incl. onestep)

7ede623

Updating the pmodel_output.rda

2ac0603

Update tests

34930dc

Fix parameter used in runread_pmodel() for onestep

dfbee4d

* runread_pmodel() had a bug for onestep. Now fixed. * Also includes changes to tests to remove warnings.

Update reference value tests

74c125f

Fix R Cmd check NOTE

7b4a3b9

Simplify runread_pmodel_f() and make more robust

ddd247c

This reduces code duplication (parallel and non-parallel versions) and makes pmap more robust by using named arguments instead of relying on positional arguments.

Further streamline runread_pmodel_f()

6f2c4be

Update ordering of variables in onestep output

b512ce6

Reduce file size of example data set for CRAN

fffe7dd

Reduce filesize of rendered vignettes for CRAN (by using some svg)

37519c3

Revert "Reduce filesize of rendered vignettes for CRAN (by using some…

a13489f

… svg)" This reverts commit 37519c3.

Reduce package file size for CRAN by removing sites and years from ex…

53c2211

…ample data

Update reference value tests

0d8f5f6

Speed up example of createMixedPrior

1b39a4d

Change output of run_pmodel_f_bysite() to tibble()

b4356ab

Make calib_sofun() backward compatible with old data format

ca4c100

Simplify calib_sofun(), handle backward compatibility ('run_model') o…

6c0a2f3

…n level of cost functions.

Remove most use of old_format drivers to remove them from package.

be4cf50

Change targets column from named list to tibble

e71b4a0

This also updates `cost_likelihood_pmodel()` and`cost_rmse_pmodel()`

fabern added 5 commits December 16, 2025 12:13

Remove old '_vcmax' leaf trait data sets

4e38cab

Refrain from using superscript minus (\u207B)

bca79c4

Fix R CMD check

1fe0ee9

Rename old cost_likelihood_pmodel to cost_likelihood_pmodel_legacy an…

7a08e0f

…d replace with new cost_likelihood_pmodel1 => cost_likelihood_pmodel_legacy cost_likelihood_pmodel2 => cost_likelihood_pmodel

Rename get_mod_obs_pmodel_bigD13C_vj_gpp => get_mod_obs_pmodel

3d5dcd6

fabern requested review from stineb and removed request for stineb December 16, 2025 12:56

fabern added 2 commits December 16, 2025 14:22

revised code for generating drivers (replaces #299)

59d28ad

Fix docs

f1987c0

fabern mentioned this pull request Dec 16, 2025

revised code for generating drivers #299

Closed

Apply styler

913ee7a

library(styler); styler::style_pkg(style = styler::tidyverse_style, strict=FALSE)

fabern added 6 commits December 17, 2025 13:28

Reduce example runtime (>5s on Windows)

e2a8c7e

Fix description in CITATION.cff

992a0b6

Rename to former use of nrChains in settings_runMCMC

2ef1ce7

n_chains_independent => nrChains n_parallel_independent => n_parallel_nrChains

Refactor and documentation cleaning

f7efadb

Make snapshot test use whc=432.375

448e0e6

fabern commented Dec 17, 2025

View reviewed changes

Revert "Make snapshot test use whc=432.375"

8bc20b9

This reverts commit 448e0e6.

fabern merged commit 3a93439 into master Dec 19, 2025
7 checks passed

fabern deleted the update-vignettes-with-gmd-paper branch December 19, 2025 09:07

fabern added a commit that referenced this pull request Jan 6, 2026

Reapply "Merge pull request #322 from geco-bern/update-vignettes-with…

6e1e2d9

…-gmd-paper" This reverts commit 4849557.

fabern mentioned this pull request Jan 6, 2026

Prepare addition/replacement of p-model example data - attempt 2 #324

Draft

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Prepare addition/replacement of p-model example data#322

Prepare addition/replacement of p-model example data#322
fabern merged 81 commits into
masterfrom
update-vignettes-with-gmd-paper

fabern commented Nov 14, 2025 •

edited

Loading

Uh oh!

stineb commented Dec 16, 2025

Uh oh!

fabern commented Dec 16, 2025

Uh oh!

fabern commented Dec 16, 2025

Uh oh!

stineb commented Dec 16, 2025

Uh oh!

fabern left a comment

Uh oh!

Uh oh!

fabern commented Jan 6, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

fabern commented Nov 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

stineb commented Dec 16, 2025

Uh oh!

fabern commented Dec 16, 2025

Uh oh!

fabern commented Dec 16, 2025

Uh oh!

stineb commented Dec 16, 2025

Uh oh!

fabern left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

fabern commented Jan 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

fabern commented Nov 14, 2025 •

edited

Loading

fabern commented Jan 6, 2026 •

edited

Loading