[OP] Address feedback for Dashboard

[tovaz92]

As per title we address here all the comments and feedback to improve v0.19 for the DESP dashboard

[codecov]

Codecov Report

❌ Patch coverage is 50.93633% with 131 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (v0.19-operational@c1c59c9). Learn more about missing BASE report.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@                 Coverage Diff                  @@
##             v0.19-operational    #2747   +/-   ##
====================================================
  Coverage                     ?   72.56%           
====================================================
  Files                        ?      122           
  Lines                        ?    11826           
  Branches                     ?     2228           
====================================================
  Hits                         ?     8582           
  Misses                       ?     2362           
  Partials                     ?      882           

[jhardenberg]

Updated figures here: https://dev-climatedt-internal-dashboard.2.rahtiapp.fi/aqua?s=(e~!climatedt-o25.1!IFS-NEMO!historical-1990!r1

[jhardenberg]

Before merging this let's remember that many of these need to be implemented in a separate PR also for the main

[oloapinivad]

I would also wait a discussion with ECMWF to be sure we do not to modify things twice

[jhardenberg]

removed tropical rainfall from the config analysis file since it is not working anyway

[jhardenberg]

I had to deactivate multilevel trends , since they plot the surface continents also at depth

[jhardenberg]

multilevel trends are back, figure fixed

[jhardenberg]

@ghossh the multilevel trends were plotting surface land/sea masks also at depth, which is misleading.
With the latest change this is now fixed, see the figure above.

There is an additional severe problem though:

Here is exactly what the multiple_maps/plot_map()routine currently does:

• For each panel:

  • It calculates vmin and vmax from the values, specifying that symmetric values are required.
  • It defines a ticks variable from vmin and vmax using linspace.
  • It then completely ignores vmin, vmax, and ticks, letting AQUA’s plot_single_maps() calculate them automatically . The problem is that by default, since sym=True is not passed to plot_single_maps() a non-symmetrical colorbar will be used.

• Finally, it displays a colorbar at the bottom which is just the colorbar of the panels in the last row, using the last defined vmin and vmax

To summarize: there is only one colorbar per variable but each depth actually uses a different range. The colorbar itself does not correspond to the actual colorbar of the panels in the last row, since it was computed with the symmetric option while the panels were plotted without.

To be done: implement a single set of symmetric vmin and vmax for each variable (one for T and one for S) , pass these vmin ad vmax to aqua.graphics.plot_single_maps() and then use the same also for the colorbars at the bottom

[ghossh]

Thanks Jost, I'm checking it now.

[oloapinivad]

Hi @ghossh I have other few comments on MLD plots

• Can we use another palette for MLD? Even viridis would be enough, but jet is not very informative.
• Can we reduce the max range from 2000 to 1500 given that none of the regions reaches the max?
• This one is more complicated I guess, but I am seeing that we are missing the areas of deep water formation because boxes are too small. We might need larger box for plotting it, but keeping the same boxes for stratification ones.
• please ask from feedback to Pablo, but the southern hemisphere deep water formation seems completely off. Are we sure that JJA is the right moment?

[ghossh]

@oloapinivad, can you please share with me the analysis results dashboard link or location? I can show it to Pablo

[oloapinivad]

Sure, please check the MLD plots for Weddel and Ross sea for [IFS-NEMO](https://dev-climatedt-internal-dashboard.2.rahtiapp.fi/aqua?s=(e~!climatedt-gen2!IFS-NEMO-5km!baseline-hist!r1~~d~!Info~Status~Climate_metrics~Radiation~Atmosphere_2D~Atmosphere_3D~Water_cycle~Teleconnections~Ocean_2D~Sea_Ice~~s~(Climate_metrics~!Gregory_plot~Performance_indices~Global_means~~Radiation~!Boxplots~Time_series~Annual_differences~Seasonal_differences~~Atmosphere_2D~!Time_series~Seasonal_cycles~Annual_differences~Seasonal_differences~Zonal_profiles~~Atmosphere_3D~!Vertical_differences~Annual_differences~Seasonal_differences~~Water_cycle~!Time_series~Seasonal_cycles~Annual_differences~Seasonal_differences~Zonal_profiles~~Teleconnections~!NAO~ENSO~~Ocean_2D~!Annual_differences~Seasonal_differences~~Sea_Ice~!Time_series~Seasonal_cycles~Differences)

[ghossh]

Hi @oloapinivad
I just talked with Pablo about the MLD plot. He is not aligned with fixed colorbar limits since EN4 shows a very different range in both cases in the North Atlantic and the Southern Ocean.

[oloapinivad]

Hi @oloapinivad I just talked with Pablo about the MLD plot. He is not aligned with fixed colorbar limits since EN4 shows a very different range in both cases in the North Atlantic and the Southern Ocean.

But we need it for comparison between models, if the model is underestimating we need to see it clearly. What about the regions seasons?

[portegam]

Hi, just to comment that JJA is the season where the MLD is deeper in the Southern Ocean, and DJF is the same for the Northern Hemisphere.

For the default figures, it might be better to show wider areas that contain the full Southern Ocean and the Artic+North Atlantic, so we can see the deep water convection regions in just two plots. I would keep 2000m as the maximum MLD in the Northern Hemisphere, and 1000m in the Southern Ocean.

[oloapinivad]

Thanks @portegam, thanks for confirming the seasons. I totally agree on both your comments concerning default regions, it would be great if this can be implemented. Ideally we could use a polar projection.

[ghossh]

Hi @jhardenberg, 2e0e0cc should stabilize vmax and vmin definitions in multilevel trends. Feel free to suggest if you see any more change is needed. Thanks!

[jhardenberg]

@oloapinivad why did you have to remove the ref list functionality in plot_timeseries?

[oloapinivad]

@oloapinivad why did you have to remove the ref list functionality in plot_timeseries?

I did not remove it. I tried to use it and then realize it was not implemented, please check the main, there is a #TODO flag https://github.com/DestinE-Climate-DT/AQUA-diagnostics/blob/d44b7d977c50987e1d69887c328945af3245d319/aqua/diagnostics/timeseries/plot_timeseries.py#L132

[jhardenberg]

climatedt-gen2/IFS-NEMO/baseline-hist in the dashboard has an updated version of this. You can seaice colormaps and lonlatprofiles changes @tovaz92 @mcadau

[jhardenberg]

@silviacaprioli it looks as if titles now say "Nino 3.4 index" but descriptions still say "ENSO index"

[jhardenberg]

@silviacaprioli false alarm... I guess old and new figures got mixed .... please ignore for now

[jhardenberg]

ok @tovaz92 do not look at the seaice figures...the diagnostic actually crashed and did not produce new figs....to be investigated

[jhardenberg]

@tovaz92 ok, seaice this time ran correctly and I see new colormaps

[jhardenberg]

A legend still says "Linear regression of sea surface temperature with ENSO index "