jbusecke
commented
3 years ago Hey @abby-baskind these look great!
Some pointers to make the code a bit sleeker (but overall there is nothing wrong with doing things manually first!):
1) Use for loops to avoid repeating code
#You dont have to spell out all the axes, you can just keep them in an array
fig, axarr = plt.subplots(...)
#loop over each key, value pair of the dictionary
ax_idx = 0 # index for your axes array
for name, ds in DICdict.items():
ax = axarr.flat[ax_idx]
...write out the plotting only once
#advance the axis index
ax_idx += 1 # this is a nifty python trick to increment a variable by one
# but there is an even more elegant way
for ax, (name, ds) in zip(axarr.flat,DICdict.items()):
... fancy plotting stuff2) Consider using xarrays plotting methods. You can rewrite this
# plot plot plot
ima = axa.pcolormesh(DICa['y'],DICa['lev'],DICa.dissic)
axa.set_xlim([-80,60])
axa.invert_yaxis()
axa.set_xlabel('Latitude')
axa.set_ylabel('Depth (m)')
cbaxes = figa.add_axes([1.03, 0.1, 0.02, 0.8])
cbara = figa.colorbar(ima, cax = cbaxes)
cbara.set_label(DICa.dissic.attrs['long_name']+' ('+DICa.dissic.attrs['units']+')', fontsize = 14)
axa.set_title(DICa.attrs['source_id']+ ' '+ DICa.attrs['table_id']+ ' '+ DICa.attrs['grid_label'])like this (I am already using the variables looped over in the forloop (e.g. axa=ax and DICa=dissic_section):
dissic_section = ds.dissic.sel(.... # select/average your section. You have to do this inside the loop aswell.
im = dissic_section.plot(x='y', y='lev', yincrease=False)
ax.set_title(DICa.attrs['source_id']+ ' '+ DICa.attrs['table_id']+ ' '+ DICa.attrs['grid_label'])xarray should handle e.g. the colorbar, units etc and might save you some time and code.
3) Set colorlimits? From your code it seems like you tried to add a colorbar, but I cant see it. To make the plots comparable, consider settting the colorlimits to the same value within the loop
im=ds.dissic.plot(..., vmin=0, vmax=5000) #not sure what a reasonable range is here4) Easier string formatting
Python has a neat way of formatting strings (f-strings), so you could replace DICc.attrs['source_id']+ ' '+ DICc.attrs['table_id']+ ' '+ DICc.attrs['grid_label'] with f"{DICc.attrs['source_id']} {DICc.attrs['table_id']} {DICc.attrs['grid_label']}"
One comment about the method:
You are currently selecting the longitudes based on the x dimension (.sel(x=slice(180,200)).mean('x', keep_attrs=True)). x is only a nominal longitude (in some models it is just a logical index - counting from 0 to however many elements x has), the safer way to select the region is based on the actual 2D longitude lon:
dissic_section = ds.dissic.where(np.logical_and(ds.lon<=200, ds.lon>=180), drop=True).mean('x', keep_attrs=True)Further down the line it might be of use to regrid each dataset to a common grid, but for now this here should suffice.
gmacgilchrist
abby-baskind
hey @jbusecke and @gmacgilchrist (graeme, ignore this while youre on vacation, it's non-urgent)
So here's some code from my dissic_plots. I'm trying to make subplots of zonal DIC, and it works (i attached an image to prove plots were in fact made), but there has to be a more efficient way to do this. Like, there has to be a more efficient way to get the intake keys instead of copy pasting them. And I bet there's a way to loop through making the slices and making the subplots. But idk how to do that, so lemme know if you know a way to make this code more efficient and less like a first-grader wrote it