How to add coastlines to a UXARRAY plot

[bstephens82]

Apologies if this is duplicating another discussion (I looked but didn't see one quite like what I was looking for), but I am just wondering how to include coastlines in uxarray plots. In some maps I've seen it's pretty obvious where the continents are simply from looking at the contours, but I have some noisier data that's unfortunately not so clear. I can plot a map, but without the coastlines it's hard to interpret. Is there a simple fix for this? Currently I'm using the command:

uxds['ug_mean_lev'][0,:].plot(title="ug_mean_lev", cmap='cubehelix_r', projection=ccrs.PlateCarree())

[philipc2]

Hi @bstephens82

Since our plotting API is written using GeoViews, you can add any feature you'd like by using the following syntax

import geoviews.feature as gf
projection = projection=ccrs.PlateCarree()

uxds['ug_mean_lev'][0,:].plot(title="ug_mean_lev", cmap='cubehelix_r', projection=ccrs.PlateCarree()) * gf.coastline(projection=projection)

You can see an example of this in our MPAS Visualization Example. I would note that only our Point plotting currently supports other projections as of right now using our direct API. We are working on getting this in soon.

If you are interested in a more classic Matplotlib implementation using Cartopy, we have recently added a new User Guide section describing how to achieve this. This implementation does support any projection (even for Polygon plotting).

Using Matplotlib, it would look something like:

import matplotlib.pyplot as plt
import cartopy.crs as ccrs
import cartopy.feature as cfeature

projection=ccrs.PlateCarree()

# create a Matplotlib PolyCollection
pc = uxds['ug_mean_lev'][0,:].to_polycollection(projection=projection)

pc.set_antialiased(False)
pc.set_cmap("cubehelix_r")

fig, ax = plt.subplots(
    1,
    1,
    figsize=(10, 5),
    facecolor="w",
    constrained_layout=True,
    subplot_kw=dict(projection=projection),
)

# add geographic features
ax.add_feature(cfeature.COASTLINE)
ax.add_feature(cfeature.BORDERS)

ax.add_collection(pc)
ax.set_global()
plt.title("ug_mean_lev")

[bstephens82]

Thanks very much @philipc2! I got your first suggestion working. Regarding your second suggestion, does that require a recent update to uxarray or something? I get error messages when I try to run it, like it complains that "projection" is not a recognized argument for to_polycollection. Thanks again.

[philipc2]

It was added recently in our v2024.07.0 release.

It's available both on PyPi and Conda Forge

[bstephens82]

Hi @philipc2 thanks again for your help so far, I've figured out those issues I think. I had one other question...what if I have a variable on an unstructured grid that's not in the uxds dataset defined above, but calculated from a variable on an unstructured grid, and I want to plot that? For example let's say I take a variable directly from the unstructured grid, but average it vertically so it's only 2D. Now I can't use the uxds['variable-name'].plot() syntax...is there another way to do that using uxarray? Thanks!

[bstephens82]

@philipc2 Just checking back to see if there was an answer for this. Also, and apologies for the noob question, but is there an easy way to plot multiple panels using uxarray? Like plotting two or more maps on a single figure? Thanks again.

[philipc2]

Hi @bstephens82

I must have missed this on my GitHub notifications! Thanks for the extra ping.

Can you try doing diff[0].plot(title="ug_mean_lev", projection=ccrs.PlateCarree()) * gf.coastline(projection=projection)

IF you can show a screenshot of what the print out for diff looks like as well that would be helpfull.

but is there an easy way to plot multiple panels using uxarray?

If you are using our direct plotting API (i.e. .plot()) , you can use the + operator between two plots to add them together. Since our plotting API is written around HoloViews, we use the same syntax that is shown here.

Here is an example of two plots stacked on top of each other.

I'd like to share our Visualization Cookbook if you haven't seen it yet. We need to put a few updates in, but it should still be a good resource.

[bstephens82]

Thanks! Using a "+" worked to get multiple plots, so that's great. Regarding taking the difference, it looks like you're saying to try removing the ",:" after the "0", and I tried that but it didn't work. The error message is the same. print(diff) gives

<xarray.UxDataArray (time: 30, n_face: 48600)>
array([[ 0., 0., 0., ..., 0., -11., -1.],
[ 0., 0., 0., ..., 0., 0., -4.],
[ 0., 0., 0., ..., 0., 0., 0.],
...,
[ 0., 0., 0., ..., -1., -3., 0.],
[ 0., 0., 0., ..., -1., -3., 0.],
[ 0., 0., 0., ..., -47., -3., 0.]], dtype=float32)
Coordinates:

• time (time) object 1979-01-01 00:30:00 ... 1979-01-01 15:00:00
  Dimensions without coordinates: n_face

[philipc2]

@bstephens82

There is a bug in our plotting API that requires each data array to have a name. I will work on fixing this. For the meantime, you can assign a name to diff using diff.name='some name'

[bstephens82]

I see, great, it's working now. I will check out Holoviews, it seems like maybe that would help a lot because I've struggled with the syntax a lot for making plots this way. I know matplotlib is another option, I will also be looking at that. One last hopefully simple question if you don't mind--if I have a 2D field and want to plot it on a map, is there an easy way to limit the coutour limits? For example the data in the plot might go from 0 to 100, is there an option to limit the contours from 25 to 50? Thanks a lot again for all your help.

[bstephens82]

Hi @philipc2, I have another small question about something that doesn't seem to be working. I would like to use the .where() functionality with a uxds array, something like

variable = variable.where(variable<0,0)

and this doesn't give any complaints by itself, but when I try to plot the result of something like this, it gives a weird complaint about not recognizing the "title" I've put into the .plot() command. Here is the code I'm trying to use:

upwp_ug = uxds['UPWP_UG_CLUBB']
upwp_ug = upwp_ug.where(upwp_ug<0,0)
#upwp_ug = abs(upwp_ug)
upwp_ug_sum=upwp_ug.sum(dim='ilev')
print(upwp_ug_sum.shape)
upwp_ug_sum.mean(dim='time').plot(title="3a1 minus dflt, sum(abs(UPWP_UG_CLUBB)) (m2/s2)", projection=ccrs.PlateCarree()) * gf.coastline(projection=projection)

If I comment out the line with .where() and uncomment the abs() value line, it plots fine. But as is, it complains about the title:

AttributeError: Line2D.set() got an unexpected keyword argument 'title'

Any ideas? Does .where() not work yet with UXARRAY? Thanks!

[philipc2]

Looks like the return type is xarray.DataArray, which is not what we'd like. I'm creating an issue for this and will look into it.

[bstephens82]

Hi @philipc2, is there an easy way to get my uxarray plots on a logarithmic color scale? I was able to figure out how to set limits to the color scale but not how to get it on a log scale. Thanks! Here's a piece of code as an example:

upwp_ug_sum.mean(dim='time').plot(title=f"sum", projection=ccrs.PlateCarree()) * gf.coastline(projection=projection)

[philipc2]

You'd like to set logz=True

Here's a good reference guide for most supported parameters!

https://hvplot.holoviz.org/user_guide/Customization.html