CODEWITHSUNDEEP
multidimensional-arraypython-xarrayfacet-grid
angilaka
angilaka

Reputation: 41

xarray dataset: Faceted plotting over data variables (data_vars)

I'm a newbie here and also to xarray.

I have xarray datasets (dimension: time,lat,lon) which contain several data variables (e.g. ModelA, ModelB) that have identical dimensions&coordinates and I want to make multipanel plots over all variables of e.g. the first timestep.

For this, I'm trying to use the xarray facet plotting utility. After trial/error, I'm getting this done, but only in an extremely inelegant/inefficient manner, and I would very much appreciate any advice how to do his better.

I particularly wonder how combine two xarray DataArrays so that they appear as two separate data variables in an new xarray DataArray/DataSet, and how to tell xarray-plot() that it should take the data variables as "col" argument for facet plotting.

Thanks a lot for any help! Regards A.

My code looks as follows:

import numpy as np
import xarray as xr
import pandas as pd
from datetime import datetime
import nc_time_axis
import cftime

#(1) == Create dummy data, dimensions and coordinates
ntimes, nlat, nlon     = 3, 8, 5
lat          = np.linspace(-10,10,nlat) 
lon          = np.linspace(-20,20,nlon) 
year, month, daystart  = 2016, 3, 1
dayend       = daystart + ntimes
datestart    = str(year).zfill(2)+'-'+str(month).zfill(2)+'-'+str(daystart).zfill(2)
cfdates      = xr.cftime_range(datestart, periods=ntimes, freq='1D', calendar='noleap')    
data         = np.random.randn(ntimes, len(lat),len(lon))   

#(2) == Create template 3dim xarray dataarray (xrda) called varA
xrda3D       = xr.DataArray(data,
                      coords=[cfdates, lat, lon], 
                      dims=['time', 'lat','lon'])   

#(3) == Create xarray dataset (xrds) with two data variables called "varA" and "varB"

xrds3D2var = xr.Dataset({'varA': xrda3D, 'varB': xrda3D*1.5})

#--> failed prior attempts, both leading to TypeError: unhashable type: 'DataArray' (?????)
#(a) xrda3D.name = "varA"; xrda2add=(xrda3D*1.5).rename("varB"); xrds3D2var = xr.Dataset(xrda3D,xrda2add)   
#(b) xrds3D2var= xr.Dataset({'varA': xrda3D, 'varB': xrda2add} )                                                    #

#(4) == create xrda with data variables encoded in additional dimensions "vardim" 
#    == like this, facet plotting can be used with name="vardim"

ovarname = '4Ddata'
arrlist = []
for varname in xrds3D2var.keys():
    tmp      = xrds3D2var[varname]         #-- select xrda of individual data variables 
    arrlist.append(tmp)                    #-- append to list
xrda4D       = xr.concat(arrlist, pd.Index(list(xrds3D2var.keys()), name='vardim')) 
xrda4Dt      = xrda4D.isel(time=0)
xrda4Dt.plot(col="vardim")                                                              

[xarray facet plot over data variables produced from code][1]


  [1]: https://i.sstatic.net/ldMuT.png

Upvotes: 4

Views: 841

Answers (1)

Ales
Ales

Reputation: 503

You should check .to_array(). Then your code would be like the following

xrds3D2var.to_array().isel(time=0).plot(col = 'variable')

Upvotes: 2

Related Questions