jbisits / oceanrasterconversions.jl Goto Github PK

The two code scripts are not that related and potentially load things that a user may not want. Reworking the scripts into modules will allow an individual part of the package, and the exported functions, to be accessed without need to have the whole (admittedly not huge..) package.

It looks like there is a PR in Makie that will allow colourscale to be set in the heatmap function but until then can do it manually by

hm = heatmap!(ax, stack_hist.histogram.edges..., log10.(stack_hist.histogram.weights);
              colorrange = (0 5),
              lowclip = :white)

This all works locally but there is an issue when building the docs. It looks like NCDatasets.jl is able to do exactly what I have done and have it work but when deploying the docs the authentication is denied. If I do not find the solution I will reach out to them.

Rename ECCO_example to conversion_example. For the demonstration of the Histogram good idea would be to produce something similar (or even demonstrate it can do the same) as xhistogram.

Currently the stable documentation is not the most recent stable version, it is some older version. I thought this would be fixed with recent updates but still need to solve what is going on.

Use DocStringExtensions (maybe ($METHODS)) and do the same for OceanVariableConversions module.

To show some of the other methods I think a good idea will be to extract a T-S profile, then convert these variables and maybe compute a density to show density with depth.

There is now more than just conversions of Raster variables so maybe need to think about renaming the package to reflect this. Instructions on how to do this if it does get renamed.

This also will close #34 as it is really just converting ocean variables. Clearly need to update README.md, docs and github repo page as well.

I will get to making documentation eventually but in the short term I will add a quasi example and how the package works into the README.

If the convert_arguments method could turn the zero entries of the Histogram.weights field into NaN would not need the colorrange and lowclip arguments in the plotting. Likely these values cannot be updated (as I think Histogram is an immutable struct) so maybe there is a way in the convert_arguments method to convert these to NaN's?

This is for consistency with the rest of the functions.

E.g. log10 to avoid the workaround for plotting I currently have. Would involve 0 -> NaN for the conversion.

@JuliaRegistrator register

@JuliaRegistrator register

Choosing a slice of the data generates a Raster but is has dims and refdims. Need to address this to be able to select a slice and then apply the conversion/computation functions.

Need to:

• add explanation of variable names and how to write them in the julia repl
• remove GibbsSeaWater dependency from the docs as this is not even used
• once package is registered remove the include and the local using .Ocea...

convert_ocean_vars uses read at the start of the function which loads the RasterStack to memory and makes things fast. Prior to adding this it took a very long time (actually did not complete before I exited) for the ECCO_example.jl to run. read might also need to be implemented in the functions that are exported and used without convert_ocean_vars.

@JuliaRegistrator register

Package now registered so can just use 'using'.

@JuliaRegistrator register()

@JuliaRegistrator register

I also have been using these a lot so would be good to add them.

@JuliaRegistrator register

Just a method to access the Histogram then using the existing normalize!(::Histogram) but will add convenience.

This issue is used to trigger TagBot; feel free to unsubscribe.

If you haven't already, you should update your TagBot.yml to include issue comment triggers.
Please see this post on Discourse for instructions and more details.

If you'd like for me to do this for you, comment TagBot fix on this issue.
I'll open a PR within a few hours, please be patient!