<!DOCTYPE html>
Using regionmask to extract data from a specific country¶
The python package regionmask makes it easy to mask a netcdf with a specific country or another region. Its main advantage is that it performs the step of rasterising a shapefile for you, and will give you a DataArray mask with the same dimensions/co-ordinates as your input dataset.
- It already contains shapefiles so you don't need to download it yourself.
- But if you want you can also use your own shapefiles.
Here is an example of extracting PM2.5 data from Russia
In [9]:
import regionmask
import xarray as xr
# replace with your netcdf
da = xr.open_dataarray('/nfs/a340/eebjs/hiadata/pm25_annual_means_sigmasked/pm25_2003_fireon.nc')
# make a plot to show what the data originally looks like
da.plot.imshow(vmax=50)
Out[9]:
<matplotlib.image.AxesImage at 0x7f825c0d5d50>
In [10]:
# extract the regionmask of your choice
# here I choose countries at 1:50m resolution
countrymask = regionmask.defined_regions.natural_earth_v5_0_0.countries_50
# create a mask from your xarray DataArray
mask = countrymask.mask(da)
mask.plot.imshow()
Out[10]:
<matplotlib.image.AxesImage at 0x7f825575b490>
This returns a mask with a different number for each country. You can lookup the number that corresponds to each country in multiple ways, e.g.
In [ ]:
print(countrymask.map_keys('GB'))
or getting a dataframe of the mask
In [17]:
df = countrymask.to_dataframe()
print(df.loc[df.names == 'United Kingdom'])
abbrevs names numbers 31 GB United Kingdom
In [18]:
# then you can use the mask to select a country
# for example if you want Russia
russia_id = countrymask.map_keys('RUS')
# then you can use this to mask your DataArray
da_masked = da.where(mask == russia_id)
da_masked.plot.imshow(vmax=50)
Out[18]:
<matplotlib.image.AxesImage at 0x7f82050b9010>
