Australian Temperature and Precipitation Change in CMIP6 SenarioMIP¶

In this notebook we demonstrate how to calculate the temperature and precipitation change in Australia using CMIP6 SenarioMIP prediction under two emission scenarios:

Use xarray to open CMIP6 data
Extract useful variables
Calculate yearly average data
Plot the temperature and precipitation change in Australia under two different future scenarios
Compare the temperature and precipitation difference between the two scenarios

This example uses Coupled Model Intercomparison Project (CMIP6) collections. For more information, please visit data catalogue and terms of use.

Authors: NCI Virtual Research Environment Team
Keywords: CMIP, Xarray, temperature, precipitation
Create Date: 2020-Apr

Load Library¶

In [1]:

import xarray as xr
%matplotlib inline

About data¶

In CMIP6 data collection, there are four future scenarios in ScenarioMIP: ssp126, ssp245, ssp370 and ssp585. See more information about ScenarioMIP scenarios

Below, we have set up these attributes in order to get the future projection data under the ssp126 and ssp585 scenarios using member 'r1i1p1f2' of CNRM_CM6-1 model simulations as an example.

Define CMIP data searching criterion¶

In [3]:

Cmip6Dir='/g/data/oi10/replicas/CMIP6'
Activity='ScenarioMIP'
Institute='CNRM-CERFACS'
Source='CNRM-CM6-1'
Experiment='' #leave it blank for later assignment
Member='r1i1p1f2'
Table='Amon'
Var=''  #leave it blank for later assignment
Grid='gr'
Version='v20190219'
Period='201501-210012'

Australian temperature change under the lowest emission scenarios¶

Use xarray to open the temperature data (tas) of the lowest emission scenario (ssp126)

In [4]:

Var='tas'
Experiment='ssp126'
ds=xr.open_dataset(''+Cmip6Dir+'/'+Activity+'/'+Institute+'/'+Source+'/'+Experiment+'/'+Member+'/'+Table+'/'+Var+'/'+Grid+'/'+Version+'/'+Var+'_'+Table+'_'+Source+'_'+Experiment+'_'+Member+'_'+Grid+'_'+Period+'.nc')

Extract temperature 'tas' data

In [5]:

tas_126=ds.tas

Calculate regional mean of Australia (lon: 115E-155E, lat: 45S-15S), and then calculate the yearly average of the monthly data.

In [6]:

tas_126_yr=tas_126.sel(lon=slice(115,155),lat=slice(-45,-15)).mean(dim=('lat','lon')).resample(time='Y').mean(dim='time')

/g/data/dk92/packages/xarray/core/common.py:1124: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.
The new arguments that you should use are 'offset' or 'origin'.

>>> df.resample(freq="3s", base=2)

becomes:

>>> df.resample(freq="3s", offset="2s")

  freq=freq, closed=closed, label=label, base=base, loffset=loffset

Plot the yearly temperature change during the time period 2019-2060

In [7]:

tas_126_yr.sel(time=slice('2019','2060')).plot(size=6)

Out[7]:

[<matplotlib.lines.Line2D at 0x7f66511e5610>]

Australian temperature change under the highest emission scenario¶

Similarly as above, the yearly temperature change under the highest emission scenario (ssp585) during the time period 2019-2060 is below:

In [8]:

Experiment='ssp585'
ds=xr.open_dataset(''+Cmip6Dir+'/'+Activity+'/'+Institute+'/'+Source+'/'+Experiment+'/'+Member+'/'+Table+'/'+Var+'/'+Grid+'/'+Version+'/'+Var+'_'+Table+'_'+Source+'_'+Experiment+'_'+Member+'_'+Grid+'_201501-210012.nc')
tas_585=ds.tas
tas_585_yr=tas_585.sel(lon=slice(115,155),lat=slice(-45,-15)).mean(dim=('lat','lon')).resample(time='Y').mean(dim='time')
tas_585_yr.sel(time=slice('2019','2060')).plot(size=6)

/g/data/dk92/packages/xarray/core/common.py:1124: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.
The new arguments that you should use are 'offset' or 'origin'.

>>> df.resample(freq="3s", base=2)

becomes:

>>> df.resample(freq="3s", offset="2s")

  freq=freq, closed=closed, label=label, base=base, loffset=loffset

Out[8]:

[<matplotlib.lines.Line2D at 0x7f66510d2b90>]

The difference of temperature change between the two emission scenarios¶

The average temperature change difference between the lowest and highest emission scenario from 2019 to 2060:

In [9]:

TasDif=tas_585_yr.mean(dim='time')-tas_126_yr.mean(dim='time')
print(TasDif)

<xarray.DataArray 'tas' ()>
array(1.4142151, dtype=float32)
Coordinates:
    height   float64 2.0

The above result shows that Australian temperature would have an additional 1.41 Celsius average warming in our lifetime span (2019-2060) under the highest emission scenario compared with the lowest emission scenario!

How about the precipitation change under the lowest emission scenario?¶

In [10]:

Var='pr'
Experiment='ssp126'
ds=xr.open_dataset(''+Cmip6Dir+'/'+Activity+'/'+Institute+'/'+Source+'/'+Experiment+'/'+Member+'/'+Table+'/'+Var+'/'+Grid+'/'+Version+'/'+Var+'_'+Table+'_'+Source+'_'+Experiment+'_'+Member+'_'+Grid+'_201501-210012.nc')
pr_126=ds.pr
pr_126_yr=pr_126.sel(lon=slice(115,155),lat=slice(-45,-15)).mean(dim=('lat','lon')).resample(time='Y').mean(dim='time')
pr_126_yr.sel(time=slice('2019','2060')).plot(size=6)

/g/data/dk92/packages/xarray/core/common.py:1124: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.
The new arguments that you should use are 'offset' or 'origin'.

>>> df.resample(freq="3s", base=2)

becomes:

>>> df.resample(freq="3s", offset="2s")

  freq=freq, closed=closed, label=label, base=base, loffset=loffset

Out[10]:

[<matplotlib.lines.Line2D at 0x7f6651068c10>]

Australian precipitation change under the highest emission scenario¶

In [11]:

Var='pr'
Experiment='ssp585'
ds=xr.open_dataset(''+Cmip6Dir+'/'+Activity+'/'+Institute+'/'+Source+'/'+Experiment+'/'+Member+'/'+Table+'/'+Var+'/'+Grid+'/'+Version+'/'+Var+'_'+Table+'_'+Source+'_'+Experiment+'_'+Member+'_'+Grid+'_201501-210012.nc')
pr_585=ds.pr
pr_585_yr=pr_585.sel(lon=slice(115,155),lat=slice(-45,-15)).mean(dim=('lat','lon')).resample(time='Y').mean(dim='time')
pr_585_yr.sel(time=slice('2019','2060')).plot(size=6)

/g/data/dk92/packages/xarray/core/common.py:1124: FutureWarning: 'base' in .resample() and in Grouper() is deprecated.
The new arguments that you should use are 'offset' or 'origin'.

>>> df.resample(freq="3s", base=2)

becomes:

>>> df.resample(freq="3s", offset="2s")

  freq=freq, closed=closed, label=label, base=base, loffset=loffset

Out[11]:

[<matplotlib.lines.Line2D at 0x7f6650fee850>]

The difference of precipitation change between the two emission scenarios¶

In [12]:

PrDif=(pr_585_yr.mean(dim='time')-pr_126_yr.mean(dim='time'))/pr_126_yr.mean(dim='time')
print(PrDif)

<xarray.DataArray 'pr' ()>
array(-0.02813326, dtype=float32)

This shows that on average Australia is becoming drier under a higher emission scenario.

Summary¶

This example shows how to use xarray to open CMIP6 dataset, extract useful variables from the dataset, calculate yearly average and do some arithmetic and plotting.

Reference¶

http://xarray.pydata.org/en/stable/