Sea ice sensitivity to GMST - Calculate from observational datasets by NParsonsMO · Pull Request #4220 · ESMValGroup/ESMValTool

NParsonsMO · 2025-10-13T15:59:33Z

Added capacity to calculate trends from observational datasets (though also keeping currently provided values, as Notz-style includes a "plausible" range that is beyond scope).

Observations are shown as squares.

Annual trend in siconc for NOAA dataset is slightly different between this and Roach's figures, 0.01677... vs 0.01692....

Annual trends for GMST were taken as one average in Roach's plots, so will not be identical.

The published plot used data up to 2018, but we only use up to 2014.

The hatching criterion for the Roach-style plot is now consistent with the published plot.

Added ability to plot two different time periods in Notz-style plot, ready for use with CMIP7 datasets.

Closes Add observations to sea ice sensitivity recipe #4095
Link to documentation:

Before you get started

☝ Create an issue to discuss what you are going to do

Checklist

It is the responsibility of the author to make sure the pull request is ready to review. The icons indicate whether the item will be subject to the 🛠 Technical or 🧪 Scientific review.

🛠 This pull request has a descriptive title
🛠 Code is written according to the code quality guidelines
🛠 Documentation is available
🛠 Tests run successfully
🛠 The list of authors is up to date
🛠 Any changed dependencies have been added or removed correctly
🛠 All checks below this pull request were successful

New or updated recipe/diagnostic

🧪 Recipe runs successfully
🧪 Recipe is well documented
🧪 Figure(s) and data look as expected from literature
🛠 Provenance information has been added

New or updated data reformatting script

🛠 Documentation is available
🛠 The dataset has been added to the CMOR check recipe
🛠 The dataset has been added to the shared data pools of DKRZ and Jasmin by the @ESMValGroup/OBS-maintainers team
🧪 Numbers and units of the data look physically meaningful

To help with the number of pull requests:

🙏 We kindly ask you to review two other open pull requests in this repository

- Removed obs from Roach-style graphic (left description for reference). - Fed in model datasets from recipe differently to allow for obs having a different variable. To do: - Calculate and plot the obs. - Print the obs trend values.

valeriupredoi · 2025-10-14T15:11:13Z

please give the PR a suitable title, when you have a minute (even if it's a draft) 😁

…-recipe-retry Attempt to fix error in Install

NParsonsMO · 2026-05-19T14:36:48Z

+
+    # Read the time constraint from the period
+    start_year, end_year = time_range.split("-")
+    time_constraint = iris.Constraint(


Clearly this can't add data, so if the observational dates fall outside the main dates specified in the pre-processor the date ranges won't be right (so when I accidentally only extracted data until 2007, it looked like the data for 1979-2007 was identical to the data for 1979-2014).

It's here because the obs values we have are valid until 2014 but data will be available until 2021, and that way round is fine, but it bothers me that if people change the dates in other ways it might be misleading...

I'd recommend adding a comment to the code to explain this, to help future readers of the code.

Added in e8e399a

NParsonsMO · 2026-05-22T14:37:39Z

      start_day: 1
      start_month: 1
-      start_year: 1979
+      start_year: &data_start 1979


Perhaps I should query the data period from the pre-processed netcdfs instead, then the titles wouldn't be at so much risk of being wrong if the cube didn't span the whole period (such as those passed through from CMEW). But I'd either have to assume all cubes were the same, or read them all...

I think it's better to hard-wire here, but have a warning/error in the diagnostic for cases where the data doesn't span the period. (Or perhaps the preprocessor throws a warning/error in those cases?)

It did not that I could see.

It did not that I could see.

OK. In which case (and reflecting an offline comment from Ed) it could be more obvious to specify the dates in the dataset specifications. (Still sticking with the hard-wiring approach.

alistairsellar

Thanks for the updates @NParsonsMO - these will make it easier to adapt for CMIP7 for the REF. Approving the science review.

@schlunma since you reviewed the first implementation of this, would you be OK to / have time to review this PR?

alistairsellar · 2026-05-27T10:58:03Z

+
+    # Read the time constraint from the period
+    start_year, end_year = time_range.split("-")
+    time_constraint = iris.Constraint(


I'd recommend adding a comment to the code to explain this, to help future readers of the code.

alistairsellar · 2026-05-27T11:04:30Z

      start_day: 1
      start_month: 1
-      start_year: 1979
+      start_year: &data_start 1979


I think it's better to hard-wire here, but have a warning/error in the diagnostic for cases where the data doesn't span the period. (Or perhaps the preprocessor throws a warning/error in those cases?)

Co-authored-by: Alistair Sellar <16133375+alistairsellar@users.noreply.github.com>

alistairsellar · 2026-05-29T11:55:17Z

      start_day: 1
      start_month: 1
-      start_year: 1979
+      start_year: &data_start 1979


It did not that I could see.

OK. In which case (and reflecting an offline comment from Ed) it could be more obvious to specify the dates in the dataset specifications. (Still sticking with the hard-wiring approach.

…-recipe-retry

NParsonsMO · 2026-05-30T20:43:46Z

OK. In which case (and reflecting an offline comment from Ed) it could be more obvious to specify the dates in the dataset specifications. (Still sticking with the hard-wiring approach.

I've changed where the years are set in 2bff0ef.

I've added a warning if not all periods are the same in 2971258.

If nobody specifies any years anywhere then the script will fail because ESMValTool will just pre-process all available data, and then when it tries to calculate the regression of OSI-SAF data over, say, HadCRUT5 data the dimensions don't match, so I've added a note to the documentation in 26b7d2f.

…-recipe-retry

schlunma

Thanks @NParsonsMO, code looks good to me! I only have a couple of minor comments.

Please run pre-commit run --all on your code, this should also fix most of the Codacy issues.

Thanks!

schlunma · 2026-06-01T08:06:43Z

+
+        # List the SIA obs
+        elif group == "siconc_obs":
+            siconc_obs.append(dataset)


There is no else here, so this will silently ignore all other variables apart from the ones you specified. Is this on purpose?

schlunma · 2026-06-01T08:16:45Z

+    regression = [
+        r
+        for _ in range(num_periods)
+        for r in regressions
+        for _ in range(num_stats)
+    ]
+
+    # Produces slope then r_value then p_value then std_err_slope, six times
+    statistic = [
+        s for _ in range(num_periods * num_regressions) for s in statistics
+    ]
+
+    # Create main column headers
+    data_columns = pd.MultiIndex.from_arrays(
+        [period, regression, statistic],
+        names=["period", "regression", "statistic"],
+    )

-    return dictionary
+    # Add equivalent of single-level columns for dataset info
+    first_columns = pd.MultiIndex.from_arrays(
+        [["", ""], ["", ""], ["label", "type"]],
+        names=["period", "regression", "statistic"],
+    )


It might be easier to set this up via pandas.MultiIndex.from_product.

schlunma · 2026-06-01T08:22:47Z

+    if "year" in cube.coords():
+        no_years = list(range(len(cube.coord("years").points)))
+    else:
+        no_years = list(range(len(cube.coord("time").points)))


I think the time coordinate will always be available, so this could potentially be simplified to

Suggested change

if "year" in cube.coords():

no_years = list(range(len(cube.coord("years").points)))

else:

no_years = list(range(len(cube.coord("time").points)))

no_years = np.arange(cube.coord("time").shape[0])

schlunma · 2026-06-01T08:32:51Z

+    if first_variable["diagnostic"] == "arctic":
+        title = (
+            "Trends in Annual Mean Temperature And September Arctic Sea Ice"
+        )
+        save_as = "September Arctic sea ice trends"
+    elif first_variable["diagnostic"] == "antarctic":
+        title = (
+            "Trends in Annual Mean Temperature And Annual Antarctic Sea Ice"
+        )
+        save_as = "Annual Antarctic sea ice trends"


There should be an else case with a proper error; otherwise your follow-up code will run into undefined variables errors for title and save_as.

schlunma · 2026-06-01T08:33:35Z

+        save_as = "September Arctic sea ice sensitivity"
+    elif first_variable["diagnostic"] == "antarctic":
+        title = "Annual Antarctic sea-ice area sensitivity\ndSIA/dGMST"
+        save_as = "Annual Antarctic sea ice sensitivity"


Same as above: add else case to make sure that all variables are defined.

In progress:

b3167f8

- Removed obs from Roach-style graphic (left description for reference). - Fed in model datasets from recipe differently to allow for obs having a different variable. To do: - Calculate and plot the obs. - Print the obs trend values.

NParsonsMO self-assigned this Oct 13, 2025

NParsonsMO linked an issue Oct 13, 2025 that may be closed by this pull request

Add observations to sea ice sensitivity recipe #4095

Open

TEMP

9c2103b

NParsonsMO added 23 commits October 17, 2025 09:06

Interim version

86c5c84

Merge branch 'main' into 4095-add-observations-to-sea-ice-sensitivity…

989bdf7

…-recipe-retry Attempt to fix error in Install

printing dictionary

5836245

Trying to fix trends

72eb211

Cross-obs values

b731cb6

fixing keys

65ff7a7

temp

c6d6335

adding one plot back in

879bf91

error finding

8deb66a

error finding

aa1a312

error finding

3347c45

retry partial replace

f7468da

error fixing

1658389

error with provenance

293720d

REMOVING provenance

a9a13a6

error with provenance

cc4301e

error with provenance

1704104

error with provenance

7fc87d2

nope, error was incorrect indentation

2f1df1b

revert to less good plt for colourbar

2be7b27

add in obs

106c8d5

looking for comparability to paper

aa74676

pre-commit changes

76c116a

NParsonsMO requested a review from alistairsellar November 24, 2025 13:59

NParsonsMO changed the title ~~In progress:~~ Sea ice sensitivity to GMST - Calculate from observational datasets Nov 24, 2025

NParsonsMO added 6 commits May 19, 2026 14:03

at to loc

5f8dd18

Missing docstring for main

008c678

Codacy stuff

fb98346

MORE Codacy stuff

deb0835

now pre-commit again

8d4b2e2

rename

a5694f1

NParsonsMO requested a review from alistairsellar May 19, 2026 14:02

NParsonsMO commented May 19, 2026

View reviewed changes

Comment thread esmvaltool/diag_scripts/seaice/seaice_sensitivity.py

missing space

9a9f67e

NParsonsMO commented May 21, 2026

View reviewed changes

Comment thread doc/sphinx/source/recipes/figures/seaice/Annual_Antarctic_sea_ice_trends.png

NParsonsMO commented May 22, 2026

View reviewed changes

Comment thread esmvaltool/recipes/recipe_seaice_sensitivity.yml

alistairsellar marked this pull request as ready for review May 27, 2026 11:06

alistairsellar approved these changes May 27, 2026

View reviewed changes

Apply suggestion from @alistairsellar

e1ce1df

Co-authored-by: Alistair Sellar <16133375+alistairsellar@users.noreply.github.com>

alistairsellar requested changes May 29, 2026

View reviewed changes

NParsonsMO added 5 commits May 29, 2026 13:14

Merge branch 'main' into 4095-add-observations-to-sea-ice-sensitivity…

5487f6a

…-recipe-retry

Change where time periods are set

2bff0ef

Querying time periods differently

2971258

Improving logging

e8e399a

pre-commit

76b3982

NParsonsMO added 2 commits May 30, 2026 22:12

documentation

26b7d2f

Merge branch 'main' into 4095-add-observations-to-sea-ice-sensitivity…

b57ed19

…-recipe-retry

NParsonsMO requested a review from alistairsellar May 30, 2026 21:27

word swap

1fc27c0

schlunma added this to the v2.15.0 milestone Jun 1, 2026

schlunma added the REF Important for the CMIP Rapid Evaluation Framework (REF) label Jun 1, 2026

schlunma reviewed Jun 1, 2026

View reviewed changes

Conversation

NParsonsMO commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Before you get started

Checklist

Uh oh!

valeriupredoi commented Oct 14, 2025

Uh oh!

NParsonsMO May 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

NParsonsMO May 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

alistairsellar left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

NParsonsMO commented May 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

schlunma left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

NParsonsMO commented Oct 13, 2025 •

edited

Loading

NParsonsMO May 19, 2026 •

edited

Loading

NParsonsMO May 22, 2026 •

edited

Loading

NParsonsMO commented May 30, 2026 •

edited

Loading