Add colormaps commonly used for plotting echograms?

[leewujung]

@LOCEANlloydizard : I think you had some thoughts about adding additional colormaps that are often used for echosounder data.

From the discussion in #230, I wonder if we should:

1. Add them to echopype and import them to use in echoshader -- the advantage of this option is that probably more people use echopype as of now than echoshader, so it's readily usable there, or
2. Add them to echoshader and also move the existing ek500 colormap to echoshader -- the advantage of this option is that everything related to viz will then be in echoshader, but people who have echopype will have to install echoshader to use these colormaps, and echoshader dependencies may be messy

Maybe there are 2 more options also:
3. Add a new tiny package (echocolors?? 😆) that just contains colormaps
4. Contribute all colormaps to cmocean -- this will also remove the need in the current echopype code to plug that into the colormap list.

I think option 4 may be the best ultimate solution, since it will then be in matplotlib, but it may take time to do that?

Also pinging @iamluan to see if you have any thoughts here.

[LOCEANlloydizard]

Hi @leewujung, yes I think tackling this would be really good! It also relates to this issue in echopype. Maybe we can have a little chat about it later and update our thoughts here afterwards!

[iamluan]

Also pinging @iamluan to see if you have any thoughts here.

I prefer option 4 as the most stable one.

[LOCEANlloydizard]

Hi @iamluan we're just checking the state of cmoceans, as it could be that it is not maintened anymore?

[leewujung]

Ok, @iamluan and @LOCEANlloydizard : I had a discussion with @ocefpaf on this and had the following conclusion:

It is probably the best (for us to manage and for users to not have to think about it) to have the colormaps in a lightweight separate package that only has colormaps, and then have it as dependencies in both echopype and echoshader, and use it via import as has been set up in echopype here and here.
@ocefpaf also pointed out this nice example for what we wanted to do: https://github.com/callumrollo/cmcrameri/tree/main, which uses the same mechanism to register the custom colormap with matplotlib.

What do you think about going these routes?

[iamluan]

Hi @iamluan we're just checking the state of cmoceans, as it could be that it is not maintened anymore?

Yeah, it looks like cmoceans has been inactive for 2 years.

So as I understand, we might have 2 options, which are adding a colormap to echopype and importing it to use in echoshader, and the other is to create a separate package like cmcrameri. I'm slightly more in favor of the first option, as it will be implemented quickly but still stable enough.
I'm not fully aware of the importance of using a specific colormap in an echogram, as mentioned in echopype docs, that ek500 can make certain features stand out. In that case, creating a separate package might make more sense.
Would another option be to directly use cmcrameri?

[leewujung]

Would another option be to directly use cmcrameri?

Since crcrameri is specifically for making Fabio Crameri's perceptually uniform colourmaps for geosciences available, it would be better to not add the echogram-specific ones as that could create confusion.

And yes there have been quite some discussions on what colormaps to use for echograms. I favor the option of creating a separate colormap package because of this. Just on top of my head there are:

• Blackwell et al. 2020. Colour maps for fisheries acoustic echograms.
  • has the related code/data in https://github.com/bas-acoustics/fsz242 which we can leverage and cite them
• Peña 2021. Full customization of color maps for fisheries acoustics: Visualizing every target.
  • code for doing histogram equalization in the appendix